Of North America

UNITED STATES DEPARTMENT OF THE INTERIOR

Harold L. Ickes, Secretary

FISH AND WILDLIFE SERVICE

Ira N. Gabrielson, Director

Fishery Bulletin 38

BIOLOGY OF THE ATLANTIC MACKEREL (Scomber scombms) OF NORTH AMERICA

Part I: Early life history, including the growth, drift, and

mortality of the egg and larval populations

By OSCAR ELTON SETTE

From FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Volume 50

,, **A

UNITED STATES GOVERNMENT PRINTING OFFICE

WASHINGTON : 1943

Kor sale by the Superintendent of Documents. U. S. Gove/nment Printing Office Wa*hington 25. D. C. -- - - - Price 25 ccnti ABSTRACT

This portion of a comprehensive study on the Atlantic mackerel (Scomber scombrus) treats of the early life history from spawning up to about the time the schooling habit develops, with emphasis on the quantitative aspects. Spawning takes place along the Atlantic coast, mostly 10 to 30 miles from shore, from Chesapeake Bay to Newfoundland, with perhaps Yio of the volume between the Chesapeake Capes and Cape Cod; Mo in the southern half of the Gulf of St. Lawrence, and negligible amounts elsewhere. Embryological development at the temperature usually encountered occupies about 1 week. The pelagic eggs are confined to a surface stratum 15-25 meters thick. Hatching at 3 mm. of length, larvae grow to 10 mm. in about 26 days, and to 50 mm. in an additional 40 days, by which length they approximate the typical form for adult mackerel, and assume the schooling habit. In 1932, it is estimated, 64,000 billion eggs were produced south of Cape Cod by a spawning population estimated at 100 million individuals. That year dominant northeasterly winds (which were abnormally strong) drifted one concentration of larvae, originating off northern New Jersey, and another concentration, originating off southern New Jersey, in a southwesterly direction, to localities abreast of Delaware Bay and Chesapeake Capes, respectively. A reversal of dominant winds, consequently of drift, returned both groups to northern New Jersey, by the 9-mm. stage of growth. Mortality during most of the developmental period was 10 to 14 percent per day, but was as high as 30 to 45 percent per day during the 8- to 10-millimeter period when fin development we s rapid. Survival from spawning of the eggs to the end of the planktonic phase of life (50 mm.) was in the order of 1 to 10 fish per million eggs spawned. This rale of survival is an abnormally low one since the fish from this spawning season were abnormally scarce in the adult populations of subsequent years. The low survival rate is ascribed to the abnormal amount of southerly drift, coupled with a general scarcity of plankton in the spring of 1932. BIOLOGY OF THE ATLANTIC MACKEREL (SCOMBER SCOMBRUS) OF NORTH AMERICA. PART 1: EARLY LIFE HISTORY, INCLUDING GROWTH, DRIFT, AND MORTALITY OF THE EGG AND LARVAL POPULATIONS

By Oscar Elton Sette, United Stales Fish and Wildlife Service

J- CONTENTS

Face l'aKe Introduction 149 Life history— Continued. Account of field work 151 Spawning habits 165 Synopsis of results 152 The egg 166 Significance of results 155 The larva 170 Life history 156 Growth 173 Reproductive age 156 Drift and migration 183 Fecundity 156 Mortality 191 Spawning grounds and spawning sea- Appendix 208 sons 158 Methods of determining size at Coast of southern New England maturity 208 and Middle Atlantic Statrs___ 158 Methods of collecting eggs and larvae. 209 Gulf of Maine 159 Enumeration of eggs and larvae 211

Coast of Nova Scotia 1 (10 Computations of catch per station . . 213 Gulf of St. Lawrence 160 Records of tow netting and catches Relative importance of the var- of 1932 219 ious spawning grounds 161 Sizes of youngest post-planktonic Numbers of eggs spawned and size of mackerel 235 spawning stock 164 Literature cited 230 INTRODUCTION

The common mackerel, Scomber scombnis, is found on both sides of the Atlantic Ocean, approximately between the 30th and 50th parallels of north latitude. Although American and European representatives are very much alike in appearance, life history, and habits, their ranges are discontinuous, so that the two populations may be regarded as separate races with no intermigration. Consistent with this view is the observation (Garstang, 1898, p. 284) that the two stocks differ in morphological characters. The American race has from colonial times been caught and marketed in large

1 volume. In the nineteenth century the annual yield occasionally reached 200,000,000 pounds. The present yield is about 60,000,000 to 80,000,000 pounds annually, of which the United States fishery takes about tliree-quarters and the Canadian fishery the remainder (Sette and Needier, 1934, p. 43).

1 The European race, too, is the object of an important commercial fishery, but appears never to have been held as high in esteem or occupied so high a rank among the commercial fishes of Europe as has Its American relative among the fishes of tills side of the Atlantic. Fishery Bulletin 38. Approved for publication May 15, 1939. 149 150 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Among the commercial fishes, the mackerel is remarkable for its spectacular changes in yield. To illustrate this, only a few records need be selected (Sette and Needier, 1934, p. 25). From 116,000,000 pounds in 1834 the United States catch dropped to 23,000,000 pounds in 1840, only to rise again to 137,000,000 pounds in 1848. From its peak of 179,000,000 in 1884, the catch dropped to 30,000,000 in 188(3, o;dy 2 years later. More recently it increased from 13,000,000 pounds in 1922 to 08,000,000 pounds in 1926. For the United States and Canada together the largest catch, 234,000,000 pounds, was landed in 1884, the lowest, 12,600,000 pounds in 1910. Although these fluctuations had profound effects both on the economic welfare of the fishermen and on the business of the fish markets, and although speculation, both popular and scientific, as to the causes of these sharp changes in returns from the fishery, has been indulged in for many years, no satisfying explanation has been forthcoming. This is not particularly surprising, for the scientific research concerning work on this species has been of desultory nature and unsuited to the solution of a problem as intricate as is presented by the fluctuations in fish populations. None- theless, from the fragmentary records then available, Bigelow and Welsh (1925, pp. 198-199) found evidence suggesting that the mackerel, like the Norwegian herring, was subject to marked inequalities in the annual success of reproduction or of survival to commercial size of the various year classes, and attributed the intermittently good and poor years of fishing to intermittently good and poor seasons of spawning or survival. This hypothesis, being the most reasonable one thus far advanced, determined the method of approach in the present investigation. Obviously, its pursuit required two basic series of observations: (1) An estimate of changes in abundance, and (2) determination of changes in age composition. Carried through a number of years, these observations should provide material for measuring the relative numerical strengths of year classes arising from each season's spawning, for tracing the influence of the annual increments afforded by each year class and their subsequent mortality on the success of the commercial fishery, and conversely for examining the influence of the commercial fishery both on the reproductive success and on the mortality. Accordingly, after some preliminary field work in 1925 at Woods Hole and Boston, Mass., in which various techniques of sampling and measuring were developed, a routine program of observations was commenced at the principal mackerel fishing ports. For the estimation of changes in abundance, pertinent details covering the landings by mackerel vessels were recorded to form the basis for computing catch per unit of effort for the determination of of mackerel fishing ; and age-composition, samples were drawn daily from each of a number of the fares landed. These basic observations began in 1926 and have continued to the present time. In addition, inquiries were pursued into the natural history and habits of the mackerel, since more adequate knowledge of these was required for interpretation of the data derived from the commercial fishery. During the 10 years, 1926 to 1935, sufficient material has accumulated to provide substantial contributions to the understanding of the life history of the mackerel, with special reference to its fluctuations in abundance; and, accordingly, a series of papers, 2 of which this is the first, is to be published. The present paper deals with features of the early life history, with particular reference to the understanding of variations in the annual replenishment of the commercial stock. It summarizes present knowledge

1 Results, of preliminary nature, previously published are to be found in Sette, 1931, 1932, 1933, and 1934. Also see Sette and Needier, 1934. BIOLOGY OF THE ATLANTIC MACKEREL 151 of the course of events from the time the eggs are spawned until the young mackerel attain the juvenile phase and closely resemble the adults in form and habits. Other deal with habits and papers in this series, now in course of preparation, (1) migrations, fluctuations in abundance of the commercial stock. (2) age and rate of growth, and (3) Acknowledgments.—The entire portion of the mackerel's life considered in this paper is passed suspended in the waters of the sea, hence as a member of the plankton community. Accordingly, the data were secured by towing fine-meshed plankton nets through the waters of the spawning grounds. A preliminary cruise in Massa- chusetts Bay was taken in 1926 on the U. S. Fisheries steamer Gannet, Captain Green- seasons 1927 to 1932 were on the S. leaf, commanding. Cruises in succeeding U. Fisheries research steamer Albatross II, Captain Carlson, commanding. In June 1932 the Albatross 77 was taken out of service and completion of that season's program was made possible by the kindness of the Woods Hole Oceanographic Institution in the ketch putting at our disposal for two cruises during June and July Atlantis, Captain MacMurray, commanding. Numerous persons assisted in the scientific work aboard ship. Of these, E. W. Bailey, Wm. C. Neville, and Herbert Ingersoll took part in many cruises. Wm. C. Herrington's suggestions contributed greatly to the development of the use of current meters to measure flow through the plankton nets. In the separation of eggs and larvae from the other planktonts, numerous persons assisted, but the major portion of the responsibility rested on Mildred Moses, whose in removal of the desired material. vigilance insured a constant level of accuracy Her performance of subsequent numerical computations was also an important contribution to the present results. To C. P. Winsor I am indebted for suggestions relating to the statistical treatment of the mortality curves. Certain tabulations and the graphs used herein were products of W. P. A. official project No. 165-14-6999. Throughout the investigation, and in all of its many phases, the constantly available encouragement and advice of Henry B. Bigelow has been invaluable. To the extent that this account proves readable, the reader may thank Lionel A. Walford whose editorial suggestions have been freely followed. ACCOUNT OF FIELD WORK

As before mentioned, when work began in 1925 it was strongly suspected that the fluctuations were due mainly to annual variations in the comparative success of survival through the larval stages (Bigelow and Welsh, 1925, pp. 198-199). Accord- at the outset of the in ingly, work on the early life history was begun investigation 1926. At that time, it was not known where most of the spawning took place or where the nursery grounds for larvae were located. The literature recorded the occasional finding of eggs in the sea south of the Gulf of St. Lawrence, but no larvae; waters off the United yet the spawning population apparently favored the southerly States coast as much as the northerly waters off the Canadian coast. Massachusetts this season Bay was a spring mackerel fishing ground well known to be visited at by in numerous ripe adult individuals, so the first search took place there. Towing of the various parts of the bay yielded large numbers of eggs, especially in that portion but waters partially enclosed by Cape Cod. Not only were the eggs abundant, numbers of larvae in various stages of development were found. 152 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Encouraged by this success in waters south of the previously known distribution Cod. Here were found in of larvae, search was in 1927 extended south of Cape eggs of As abundance from the offing of Cape Cod nearly to the mouth Chesapeake Bay. in Massachusetts Bay, larvae were present in abundance also. To determine whether this was the usual condition, the survey was repeated in a the same conditions were single cruise during May of 1928, when approximately found. These three seasons of prospecting for mackerel eggs and larvae completely al- in the northwest tered the previous notion that spawning was more successful portions obtained from Massa- of the range of the species. Not only were specimens regularly chusetts Bay to Chesapeake Bay, but the numbers of individuals per tow were greatly the in excess of those taken by similar methods in the Gulf of St. Lawrence during was far Canadian Fisheries expedition of 1914-15. Evidently this southerly region one in which to more important than previously supposed, and hence a suitable study variations in the survival rate during early stages. of the season However, it was still necessary to determine the length spawning successive and the duration of the period of larval development. For this purpose, of These cruises were made during the spring and early summer months 1929. proved in that in the area between Cape Cod and Cape Hatteras spawning began early April, and larval development had nearly run its course by the end of July. season were re- In 1930 and 1931, such successive cruises during the spawning methods of the abund- peated and every opportunity was taken to devise estimating ance of the various young stages. determination of verti- This development of quantitative technique required the of the incu- cal distribution so that the proper levels would be fished; determination to include bation and growth rates so that cruises might be planned at proper intervals a reliable method of meas- all the important events; and finally, it required devising nets so that hauls would be com- uring the amount of water strained by the tow 1932 and parable from time to time and place to place. By knowledge techniques to were advanced sufficiently for the survey of that season provide adequately quan- with titative data for the more important sections of this report dealing growth, close of this the Albatross II was withdrawn drift, and mortality. Toward the season, the from service as a Government economy measure. This prevented continuing of and its research into its next phase, that is, the measurement mortality accompany- of to see how is ing hydrobiological conditions through a series seasons, mortality survival contrasted with other affected by particular conditions in seasons of good conditions in seasons of poor survival. Since the hoped-for resumption of surveys are now has not yet been possible, the present available results reported. SYNOPSIS OF RESULTS

Most mackerel reach reproductive maturity when 2 years old. Some precocious others of both sexes a individuals, usually males, first spawn a season earlier and year than the males. later. The percentage of the latter is higher among the females Mackerel are said to spawn 360,000 to 450,000 eggs in a season, but this is a point fewer and indi- needing further study. Doubtless smaller individuals spawn larger successive it is not viduals more eggs than this. The eggs are ripened in batches; known how many batches or what interval of time intervenes between their discharge. of the Spawning takes place over nearly the entire spring and summer range species , is from off Chesapeake Bay to Newfoundland. By far the most important ground BIOLOGY OF THE ATLANTIC MACKEREL 153 between the Chesapeake capes and Cape Cod; second in importance, with perhaps one-tenth as much spawning, is the southern half of the Gulf of St. Lawrence. Other stretches of the coastal waters may at times receive negligible amounts of spawn, but it is safe to say that the entire Gulf of Maine (excepting Cape Cod Bay), and the entire outer coast of Nova Scotia, the northern two-thirds of the Gulf of St. Lawrence and the waters around Newfoundland are not regular spawning grounds of any importance. 154 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Embryological development is similar to that of other teleost fishes. It pro- in 2 at 21° C. gresses more rapidly in warm water than in cold, eggs hatching clays the (70° F.) and in 8^ days at 10° C. (50° F.). The prevailing temperatures on spawning grounds at the height of the spawning season are between 9° and 12° C, so that in nature the incubation period usually occupies about a week. During incubation the eggs are suspended in the sea water between its surface and the thermocline, which is usually 15 to 25 meters (8 to 13 fathoms) deep in the area studied. They have a tendency to sink gradually as development proceeds, so that the late stages are found at deeper levels than the early ones, but even so. not below the thermocline. After hatching, the young mackerel passes through three phases of development, conveniently designated as yolk sac, larval, and post-larval stages. During the — is and yolk sac stage—a matter of about 5 days the fish about 3 mm. (K-inch) long subsists on the yolk. During this period, the mouth and digestive organs develop into usefulness and the yolk sac is absorbed. During the period occupied by the and of which larval stage, that is, between yolk sac absorption development fins, lasts about 26 days, the fish grows from a length of 4 mm. (%-inch) to 10 mm. OS- the inch) in length. Then, when the fins have appeared, post-larval stage begins. It continues about 40 days and during this time the fish grows to a length of about 50 mm. Toward the end of this stage, while growing from 30 to 50 mm., the body assumes the trim fusiform shape of the adult. At that time, the fins, relative to the includes body, are even larger than in the adult, and the coloration shiny, silvery of the adult. iridescence, though still lacking the characteristic wavy black bands the During the yolk sac stage, movements are feeble, not even serving to keep the larval and are exer- fish right side up. Swimming faculties increase during stage cised in performing vertical diurnal migrations, the larvae ascending toward the do not surface at night and descending toward the thermocline at day. But they this instead drift with the water swim any considerable distances during stage ; they masses in which they are suspended. In post-larval stages, time swimming takes to the drift place, the young fish at times moving in a direction opposite prevailing of water. The schooling habit probably begins to assert itself toward the end of the adults. this stage and thereafter is followed in much the same fashion as by the main In 1932, the larvae were drifted initially in a southwesterly direction, and from body was transported about 80 miles down the coast, one subgroup drifting from the the offing of northern New Jersey to the offing of Delaware Bay; another, a reversal of offing of southern New Jersey nearly to the Chesapeake capes. Then, drift returned both groups to the offing of northern New Jersey by the time they had reached the end of the larval stage, and were 9 mm. long. The southwesterly the drift, coincided with the predominance of northeasterly winds, and northeasterly return with a reversal of dominant winds. wind Compared with other seasons, 1932 had an abnormally large northeasterly component, which left the 9-mm. larvae farther to the southwest and farther offshore than in other seasons. After the post-larval stage of active swimming commenced, the direction of travel was toward southern New England, and by the latter Shoals part of July, some of the largest of the post-larvae had even passed Nantucket and were taken off Cape Cod. In 1932 the mortality over most of the developmental period was 10 to 14 percent per day. There was a notably higher mortality of 30 to 45 percent per day during from the 8- to 10-mm. period, when fin development was rapid. Other departures BIOLOGY OF THE ATLANTIC MACKEREL 155

I he general rate, of doubtful significance, were during egg stages, when about ^ percent per day was indicated, and during the yolk sac stage (3-mm. larvae), when about 23 percent per day was suffered. The indicated total mortality, from the spawning of the eggs to the end of plank-

Ionic existence (50 mm. or 2 inches long), was 99.9996 percent. That is, the survival was in the order of magnitude of only 1 to 10 fish per 1,000,000 of newly spawned eggs. This mortality was not due to sharply higher death rate at the yolk-sac stage— a theory of year-class failure holding favor among fishery biologists. Mortality was substantial in all stages. It was greatest during fin development in the transition phase from larval to post-larval stages. The higher mortality at this time appears to have been connected with the particular pattern of drift caused by the dominant wind movement, which in 1932 left the larvae farther than usual from their nursery grounds along the southern New England coast. This, together with a general scarcity of plankton, is considered the cause of failure of the 1932 year class. SIGNIFICANCE OF RESULTS

Most conservationists lay particular stress on the maintenance of adequate spawning reserves. It is important to do so. If an annual commercial crop is to be constantly obtained, the spawning stock must be kept large enough to produce as many young as are needed to replace the fish caught by man and other predators. This can be done, in most cases, only by controlling the annual 3'ield. From this springs an obvious, but not universally appreciated, fact that accumulating a surplus of spawners is a wasteful practice, for it means holding the annual yield below the amount that the resource is capable of producing. It would be simple, for instance, to insure an adequate spawning reserve by allowing no fish to be caught. But this would be more futile than to allow all to be caught. The latter would utilize one crop, the former none. Obviously, efficient exploitation calls for an intermediate course of action, one that would permit taking the maximum annual yield commensurate with the maintenance of an adequate spawning reserve; no more and no less. But what is an adequate spawning reserve? It can be defined as one large enough to reproduce the young needed to recruit the commercial stock. Its determination is a matter of observing the numbers of recruits produced by spawning stocks of different sizes. Thus, the answer rests on knowledge of recruitment. Two things affect recruitment: First, the numbers of spawners; second, the mortality in young stages—"infant mortality." The latter is tremendous and variable. Its variability is so great that it could readily obscure such correlation between number of spawners and number of recruits as might be present intrinsically. For example, under a given quality of survival conditions a large spawning population may produce a large number of recruits and a small population a small number of recruits, but with variable survival conditions a large number of spawners might produce only a small number of recruits if infant mortality be relatively high; and conversely, a small number of spawners might produce a large number of recruits if infant mortality be relatively low. As long as one can observe only the changes in numbers of spawners and numbers of recruits, the relation between the two cannot be seen, for it is obscured by the intervening infant mortality. Therefore, as long as the effect of infant mortality is unknown, so long will the size of an adequate spawning reserve be unknown. Thus the measurement of infant mortality is the key to the problem. In the course of this study, a technique for making this measurement has been devised, and 156 FISHEKY BULLETIN OF THE FISH AND WILDLIFE SERVICE was applied during the season of 1932. With similar observations in enough additional seasons, it should be possible to determine what recruitment can be expected from given sizes of spawning stocks for particular infant mortality rates. Thus there will be determined an adequate spawning reserve, for it will be one that produces the needed average recruitment over the observed range of infant mortality rates. LIFE HISTORY REPRODUCTIVE AGE

According to information formerly available (Bigelow and Welsh, 1925, p. 205), "Some few females ripen when still not more than 11 inches long; most of them, and all males, at 12 to 13 inches." Present observations indicate first attainment of maturity at somewhat larger sizes, the difference possibly being due to the manner of measurement. The lengths given below were from snout to tip of the middle rays of the caudal fin, whereas the earlier measurements may have excluded the caudal fin. Of 1,116 mackerel sampled from catches of traps in the vicinity of Woods Hole, Mass., and at three localities on the shores of Massachusetts Bay between June 24 and July 21, 1925, the smallest male with mature gonads was 26 cm. (10M inches) long and the smallest female 29.5 cm. (UK inches). At 30.5 cm. (12 inches) 30 percent of the males and a negligible percentage of females were mature. At 34 cm. (13H inches) about two-thirds of the males and one-half of the females were mature; and at 37 cm. (14% inches) nine-tenths of both sexes were mature. (See fig. 2.) It is possible that our data may not be typical because they were taken somewhat after the peak of spawning, which usually falls in May and June, and some individuals which hadjspawned early, and whose gonads had somewhat recovered, might have been mistaken for immature individuals. The number so mistaken cannot have been large for there was little difficulty in recognizing the two categories, "ripe" and "spent," which make up our class of "mature." The mistakes, if any, because the spawning of some individuals was too long past, should have been mostly among the larger sizes, because they are usually first to appear along the coast and presumably the earliest to spawn. But among these (52 specimens over 38 cm. hi length were examined) only 1 individual appeared immature, hence the error, if any, must have been small. By means of size and age relations to be published in another paper of this series, it may be concluded that only a few males, and even fewer females, spawn as yearlings. Four-fifths of the males and two-thirds of the females spawn when 2 years old, and virtually all of both sexes when 3 years old. FECUNDITY

Various statements have appeared in the literature purporting to give the numbers of eggs spawned by individual mackerel. Brice (1898, p. 212) in "The Manual of Fish Culture" states that the average number of eggs at one stripping is about 40,000, that a 1% pound fish gave 546,000, and that the largest fish yielded probably a full 1 ,000,000 eggs. Bigelow and Welsh (1925, p. 208) say, "Mackerel is a moderately prolific fish, females of medium size producing 360,000 to 450,000 eggs, but only a small part of these (40,000 to 50,000 on the average) are spawned at any one time." But Moore, whose report appears to be based on more intensive study than others, more cautiously states (J. P. Moore, 1899, p. 5) "seldom 50,000 and frequently a much lesser number of ova are produced at one time, but the aggregate number matured (in a spawning season) in one female of average size is several hundred thousand." This is probably as precise a statement as is warranted at the present time. BIOLOGY OF THE ATLANTIC MACKEREL J 57

Moore (loc. fit.) has shown that there are successive batches of eggs ripened by an individual female during the course of the season. This introduces the uncertainty as to whether any particular enumeration has included, on the one hand, all batches destined to be spawned dining the ciurent season and, on the other hand, none that were destined to be spawned during a following season. The difficulty of making a correct decision is amply portrayed by the thorough study by Clark (1934) on the California sardine, Sardinops caerulca, a species which, like the mackerel, spawns successive batches. Clearly this subject requires additional study to provide statisti-

80 a 6

<60 x Ul K £40 40

TWO 20 YEARLINGS-P*WYEAR =f" 20 OLDS

20 22 26 28 30 32 34 36 40 OVER LENGTH, CENTIMETERS 40

10 8 9 ! I 12 13 14 LENGTH, INCHES

Figure 2.—Length and age at which mackerel reach reproductive maturity. The upper panel shows, by 2H cm. length groups, the percentage of each sex matured. The lower part shows by half cm. length groups, the numbers examined for determination of percentage of maturity. cally adequate data, and deserves such study because the ability to compute the number of eggs that can be produced by a population of known size-composition or, conversely, to compute the size of a parent population of known size-composition from the known numbers of eggs found in a spawning area would provide useful, if not indis- pensable, data for elucidating several perplexing problems connected with the fluctuations of fish populations and the management of fish resources. See pages 164 and 165 for an example of the uses of such data. 1 58 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE SPAWNING GROUNDS AND SPAWNING SEASONS

Bigelow and Welsh in 1925 (pp. 206-208) summarized the information available on the spawning of the mackerel. Apart from the generalization that mackerel spawn along the American Atlantic coast from Cape Hatteras to the Gulf of St. Lawrence mainly in spring and early summer, most of the conclusions reached at that time are " * * * now subject to revision. Their statement (p. 206) a much greater production of mackerel eggs takes place east and north than west and south of Cape Cod, with the Gulf of St. Lawrence far the most productive nursery for this fish," is particularly at variance with present available facts, as will appear from the following account, of the numbers of mackerel eggs found in the various parts of the spawning range.

Coast of the Southern New England and Middle Atlantic States

Numbers and distribution.—Until the present investigations there was little known about the spawning in the great bight bordered by the shores of southern New England and the Middle Atlantic States. Although ripe individuals are commonly taken in thp fishery in this area, no appraisal had been made of the egg concentrations to be found there; nor was it known whether larvae hatched from such eggs as were spawned there could survive; in fact it was suspected that reproduction was unsuccessful, for no larvae of the mackerel had been captured there. As a result of information gained from the surveys of the present investigation during the seasons 1927-32, this region now appears to contain the most importaut spawning grounds of the mackerel. In horizontal tows at the surface, i. e., in the stratum of densest concentration, a meter net has taken, in 20 minutes, as many as 3 185,000 eggs. In 1929 the average catch per positive tow of this kind was 2,600 eggs during the cruise of May 10 to 18, and 5,000 eggs during the cruise of May 28 to 31. These numbers may be taken as fairly typical of concentrations at the surface when and where spawning is active, and will be useful for comparison with other regions where similar data are available. More informative, in the absolute sense, are the results of oblique tows of 1932, which sampled all levels and covered syste- matically the entire region between Cape Cod and the Chesapeake Capes. The average catch of such tows, including all between May 2 and June 21, i. e., the major portion of the spawning season, and including both positive and negative tows, was slightly over 1,100 eggs. Since these tows strained 17 cubic meters of water per meter of depth fished, the average concentration was 65 eggs per square meter of sea surface. Within this region eggs have been consistently most abundant along the inner portions of the continental shelf. The area of densest distribution occupies about the inner half of the shelf off New York with the zone narrowing and trending somewhat offshore southerly, and also narrowing but trending inshore northeasterly. By far the greatest concentrations have been found regularly somewhat southerly of the Fire Island Lightship, and this undoubtedly marks the usual center of greatest spawning activity. So far as is now known, no spawning takes place in the enclosed waters of the bays and sounds west and south of Block Island. A few eggs are spawned in the southern part of Buzzards Bay and Vineyard Sound, but these are negligible in quantity compared with the spawning in open waters.

J Positive here indicates a tow in which mackerel eggs were caught. BIOLOGY OF THE ATLANTIC MACKEREL 159

Season.—Spawning begins in the southern end of this region during the middle of April about as soon as the mackerel appear in the offing of Chesapeake Bay. Thence it proceeds northeastward along the coast, taking place during the month of May off the New Jersey and New York coasts and extending into June off southern Massachusetts. In 1932, spawning in this region reached its climax about the middle seasons indicate that this is the of May. (See table 5.) Surveys of other spawning usual time of maximum spawning. Temperature at spavming.—In this region we have found mackerel eggs in water as cool as 7.3° C. (45° F.) and as warm as 17.6° C. (64° F.). In 1932, the greatest numbers of eggs (98 percent) were found in water of 9.0° to 13.5° C. (48° to 57° F.) and this may be regarded as the range in which the bulk of mackerel eggs are usually spawned in this region. Gulf of Maine

Numbers and distribution.—On visits to the western portions of the Gulf of Maine during the present investigation, eggs were found only in Cape Cod Bay. There the concentration was only slightly less than in waters south of Cape Cod of and the coast but practically none were found in waters off the outer face Cape Cod between Boston and Cape Elizabeth. Moore (1899) found them in the outer por- of in but the numbers were few. and Welsh (1925 tions Casco Bay 1897, Bigelow ; of the Gulf of Maine. The maximum p. 206) occasionally found a few in various parts haul was recorded by them as "200 plus." Although Bigelow and Welsh (1925, p. 207) say, "That Nantucket Shoals, Georges Bank, and Browns Bank, like the Scotian banks to the east, are also the fish there sites of a great production of mackerel eggs is proven by the ripe caught * * * these banks around the of the Gulf ", it now hardly appears likely that periphery of Maine can be the site of important spawnings. The records of eggs taken by Bigelow and Welsh did not include any from these banks and during the present the investigation the waters about Nantucket Shoals were visited repeatedly, and western half of Georges Bank occasionally, without finding more than negligible numbers there. It is likely that the ripe fish caught on these grounds were a part of schools destined to spawn elsewhere, presumably the Gulf of St. Lawrence, and were taken during the course of migration to that area. This is in harmony with the results of investigations on migration which are to be reported on in another paper of this series. Thus it appears that the only spawning ground regularly important in the Gulf of Maine is Cape Cod Bay. This body of water is so small compared with the grounds south of Cape Cod or with those of the Gulf of St. Lawrence that reproduction in the Gulf of Maine must be negligible compared with that of the other spawning regions. Season.—Spawning probably takes place somewhat later in the Gulf of Maine than south of Cape Cod in consequence of later vernal wanning and later incursion of mackerel into the waters of this region. It evidently was on the increase and perhaps near its maximum in Massachusetts Bay between June 9 and June 14 of 1926, when hauls taken on a hue of three stations running out from Wood End Light toward the middle of Cape Cod Bay averaged 700 and 1,200 per tow on June 9 and of the time of maximum 14, respectively. A more precise determination spawning awaits the sorting of additional hauls made in 1926 and 1930.

L I B A R R Y } =o 160 FISHERY BULLETIN OF THE FISH AND WILDLIFE SEEVICE

Coast of Nova Scotia

Numbers and distribution.-—Information on the occurrence of mackerel eggs along the coast of Nova Scotia is limited to the results of a survey in 1922 reported by 4 Sparks (1929, pp. 443-452) . Stations were occupied along the entire coast from Cape Sable to the Straits of Canso during the period May 31 to September 18, but no eggs were taken after the middle of July. For the most part the hauls yielded very few eggs, the average number taken being 14 per station, which presumably represents the sum of three tows. 6 Although Sparks stated neither the dimensions of his nets nor the duration of his tows, it may be presumed that at least the surface net was a meter in diameter at the mouth and that the tows were 15 to 30 minutes in duration.

If so, the egg concentration was exceedingly low compared with the other regions. Furthermore, the occurrence of eggs even in this low concentration was limited to a relatively narrow band along the coast (table 1). Thus the waters along the Nova Scotian coast are poorer in mackerel eggs than any others within the known habitat of the. species. Season.—Spawning occurs along the Nova Scotian coast from about the last of May to the middle of June.

Table 1.- -Number of mackerel eggs taken per station in Nova Scotian waters at variovs distances from shore

Station BIOLOGY OF THE ATLANTIC MACKEREL 161 place occasionally on the western coast of Newfoundland, but probably only in bays to 10° it is of occurrence in which the water warms up C. (50° F.) ; perhaps irregular and it is certainly of minor importance.' Season.—In the southern half of the Gulf of St. Lawrence, eggs were present as taken early as May 29 and as late as August 12. The maximum catches were on of the season was June 30, July 7, and July 8, and it may be presumed that the height in the latter part of June and early part of July. Relative Importance of the Several Spawning Regions

Because it is important to know which grounds are mainly responsible for recruitment of the mackerel population, an appraisal of the relative amounts of spawning in the four regions will be attempted, even though the available information is not adequate for precise treatment. Since these four regions are roughly equal in size and each is sufficiently large to constitute a major spawning area, it will suffice to examine only average concentration of eggs in each region. The pertinent data, in terms of average or usual number of eggs taken per positive surface tow with a meter net are as follows:

Continental shelf between Cape Cod and Cape Hatteras 3,000 to 5,000 Gulf of St. Lawrence About 300. Gulf of Maine (exclusive of Cape Cod Bay) Less than 100. Coast of Nova Scotia About 14.

Of course, these numbers cannot be taken at their face values for there are many factors affecting their comparability. However, the last two items in the list are so low that it may be concluded that the coast of Nova Scotia and the Gulf of Maine are of negligible importance as mackerel spawning areas. On the other hand, the Gulf of St. Lawrence and the continental shell' between Cape Cod and Cape Hatteras are both grounds of evidently some importance, and their comparison with each other deserves more careful consideration. The two things that might affect most obviously the comparability of the data on them are: the that. (1) the technique of towing, including the distribution of stations, (2) fact the Gulf of St. Lawrence survey took place more than a decade earlier than the tow- netting over the continental shelf between Cape Cod and Cape Hatteras. The techniques employed in the Gulf of St. Lawrence by the Canadian Fisheries Expedition obviously were not intended for quantitative purposes. According to a rule Dannevig (1919, p. 3) "The duration of the surface hauls varied somewhat, as * * *" between ten and fifteen minutes; and Huntsman (1919, p. 407) states, "The tow hauls (as distinguished from the vertical hauls) are the most unreliable, owing to lack of information in the records as to the manner in which they were taken * * *. The tow hauls were taken in a great variety of ways." Further, Hunts- man's table (loc. cit., p. 419) of hauls by the C. G. S. No. S3, which contributed most of the mackerel eggs, shows that some of these hauls in reality were oblique and that towing periods varied between 5 and 20 minutes, with the time not given for certain of the hauls containing important numbers of mackerel eggs. Furthermore, the stations were closely spaced in some portions of the Gulf and widely spaced in others. They may have chanced to be concentrated where the eggs were thickest or the contrary. Similarly, the distribution with respect to time may have been favorable to the taking of abnormally large numbers of eggs, or the contrary. On the other hand, the coverage, both as to space and time, was far from haphazard. The Princess occupied stations in the Gulf of St. Lawrence during June 9 to June 15 162 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE and again during August 3 to 12, and, in the meantime, No. S3 was making net hauls in the southern half of the Gulf during June, July, and August, the two boats together making about 50 net hauls in the productive southern half of the Gidf during the mackerel spawning season (Dannevig, 1919, charts and tables). While it cannot be said whether more intensive work over a more uniform pattern of stations would have revealed substantially a greater or less number of eggs than was taken by the Canadian Fisheries Expedition, the fact remains that only one of their hauls yielded more than a thousand eggs and only a few, more than a hundred. Experience in the area between Cape Cod and Cape Hatteras indicates that a similar coverage, with similar techniques, would have resulted in many more hauls containing thousands of eggs, and the conclusion appears inescapable that eggs were much less abundant in the Gulf of St. Lawrence in 1915 than in the area between Cape Cod and Cape Hatteras during 1927 to 1932. It is difficult to determine how much the decade of difference in the time that the Gulf of St. Lawrence and the area between Cape Cod and Cape Hatteras were investigated affects the comparability of the data on egg numbers, but at least two obvious features may be considered—annual fluctuations and long-term trends in volume of spawning. In the area between Cape Cod and Cape Hatteras the numbers of eggs were consistently high during the years 1927 and 1932. Though the methods of towing varied too much and the coverage in some years was too deficient to permit mathematical demonstration of this, in every year the eggs were sufficiently abundant to be taken by the several thousand per surface tow at favorable times and in favorable places; and it may be concluded that annual fluctuations were not sufficient to alter the general magnitude of egg production. It appears also that the numbers of spawn- era, judging from catch statistics, did not fluctuate by orders of magnitude during this period. Thus, experience suggests that the egg yield does not fluctuate markedly as long as the number of spawners does not. Referring now to the catch statistics in the Canadian and the United States fisheries (Sette and Needier, 1934, p. 43) it appears that the trend in Canada was nearly horizontal between 1915 and the late 1920's, but that in the United States the general level was about three times as high in 1929 as in 1915. If it may be assumed that the spawners are, in general, proportional to the catch and that the numbers of eggs are proportional to the number of spawners, both of which are admittedly questionable premises, then it could be argued that the 1915 Canadian data on eggs would roughly hold for recent times and the comparison justified as indicating relative amounts of spawning in the two areas in recent times. On the other hand, comparison as of 1915 might be expected to reduce by two-thirds the numbers of eggs in the Cape Cod to Cape Hatteras area, and thus indicate relatively greater importance for the Gulf of St. Lawrence. Even so, the change would not be one of order of magnitude. All available information considered, it appears most likely that the spawning in the area between Cape Cod and Cape Hatteras is distinctly more important than hi the Gulf of St. Lawrence, and though it is possible that the difference is one of an order of magnitude, with eggs so concentrated in the Cape Cod to Cape Hatteras legion as to be available in the thousands per tow, and so scarce in the Gulf of St. Lawrence as to be available in the hundreds per tow, it is also possible that the true divergence is less marked and that the numbers are really in the upper and lower levels of the same order of magnitude. The diagrammatic representation of relative egg numbers in the various regions given in figure 3 shoidd be considered with this reservation. Although the collection of more adequate data on the subject is greatly BIOLOGY OF THE ATLANTIC MACKEREL 163 to be desired, present information supports the view that the present survey has covered the most important spawning ground. The existence of large regions with little spawning near the middle of the spawning range of the species is a peculiarity that may be explained by hyclrographic conditions. It will be noted from the diagrammatic representation of relative intensity of spawning in figure 3 that the regions of greatest intensity are the southern and northern quarters of the spawning range. That of the least intensity is the middle balf of the range. between Cod The places of intense spawning, that is, the great oceanic bight Cape

Figure 3.— Relative intensity of mackerel spawning in various regions along the Atlantic coast of North America, as indicated by the average number of eggs caught in plankton nets. and Cape Hatteras, Cape Cod Bay, and the southern half of the Gulf of St. Lawrence, have this in common: they are all bodies of relatively shoal water overlying relatively flat bottom, where topography and circulation favor vertical stability, and vernal warming of the upper strata proceeds rapidly, producing temperatures suitable for mackerel spawning earlier than in the intervening areas. On the other hand, the 525293—44 2 164 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE places of least intense spawning are areas with broken bottom where tidal and general circulation produce extensive vertical turbulence, drawing cold water from the depths to the surface, thereby delaying the vernal warming of the upper strata, as a rule, until the mackerel spawning season is over. As nearly as may be determined from the information on hydrographic conditions (Bjerkan, 1919, pp. 379-403, Bigelow, 1928, pp. 550-585) and on spawning times and places (see above), the dividing line between good and poor spawning areas may be drawn at a vernal temperature of about 8° C, (46° F.). The areas that receive little or no spawn are, during the spawning season, usually colder, and those that receive much spawn are usually warmer than this temperature. NUMBER OF EGGS SPAWNED AND SIZE OF SPAWNING STOCK

A rough estimate of the total number of eggs spawned in the region between Cape Cod and Cape Hatteras can be made from the data of the 1932 survey of spawning. The average catch during the fust seven cruises was about 1,000 eggs per 17 square meters of sea surface (table 19), or an equivalent of about 200 million eggs per square nautical mile. Taking 25,000 square miles as the areas surveyed, this would amount to a total of 5,000 billion eggs. Since tbis figure is based on the average concentration during a 50-day period, and since the period of incubation would average about 7 days at the prevailing temperature, there must have been about 7 renewals or approximately 35,000 billion eggs spawned to maintain this average concentration. From a curve of numbers of eggs taken in successive cruises, it appears that perhaps one-seventh should be added to allow for the fact that the cruises did not begin early enough or extend late enough to include all the spawning. This raises the figure to 40,000 billions eggs. These are in all stages, and it may be computed from mortality rates of eggs (table 7) that this would be equivalent to 1.6 times as many newly-spawned eggs. Applying this factor, the final estimate of eggs spawned in this area in 1932 becomes about 64,000 billion. It is difficult to appraise the reliability of this estimate because of the uncertainty of its components. Judging these as well as may be, it appears that at best it may be within 25 percent of the true value and at worst only witbin the true order of magnitude. But this is only personal judgment, and since it is impossible to study statistical probabilities, there is utility in testing the result by deriving a related statistic from an entirely different source. During 1932 the catch of mackerel on or near spawning grounds during the

spawning season; that is, in area XXIII (Fiedler, Manning, and Johnson 1934, p. 96), and in area XXII, west of Nantucket Shoals during April, May, and June, was about 13,000,000 pounds. From unpublished records on size composition of this catch, it appears that about 10,000,000 pounds of it consisted of fish of spawning size, and that their average weight was nearly 1.9 pounds. Thus, a take of about 5,000,000 spawners is indicated. To estimate from this the size of the spawning stock it is necessary to know what percentage this was of the spawning stock in 1932. This may be done only in an indirect manner. The 1923 class of mackerel, after reaching spawning age, declined at a rate of 20 percent per year as measured by the catch per purse seine boat during the four seasons, 1928 to 1931 (Sette, 1933, p. 17). Tlus decline was so steady that it probably should be ascribed to mortality rather than to other causes, such as changes in availability. Of course one cannot be sure that the spawning population in 1932 was subject to the same mortality as the 1923 class during the previous years, but BIOLOGY OF THE ATLANTIC MACKEREL 165 as far as the intensity of fishing is concerned, there was no significant change between 1931 and 1932. The fleet numbered 112 seiners in 1931 (Fiedler, 1932, p. 211) and 114 in 1932 (Fiedler, Manning, and Johnson 1934, p. 97). Views may differ as to the relative part played by catch mortality and by natural mortality in causing total mortality, but by taking divergent views, say three-quarters catch mortality on the one hand and one-quarter catch mortality on the other hand, one would arrive at 15 and 5 percent, respectively, as catch mortality; or, taking a middle ground, it would be 10 percent. Similarly divergent views may be taken as to the fraction of annual mortality suffered during the spawning season. Perhaps three- quarters and one-quarter, respectively, may reasonably be taken as the extremes and one-half (or 10 percent) as the middle ground. These would give as extremes 11 and 1.25 percent that the catch during the spawning season was of the total spawning 6tock. The middle view would be 5 percent. This results in an estimated total population between 45,000,000 and 400,000,000, with a middle ground estimate at 100,000,000 individuals in the spawning population on the spawning grounds as derived from catch statistics. It will now be recalled that the estimate derived from tow net hauls was 64,000 billion eggs spawned, and if 400,000 eggs are produced by the average female (p. 156) the indicated spawning population would be 160,000,000 females, or 320,000,000 fish of both sexes. This is within the extremes computed from the catch data and about halfway between the middle and largest figures. Considering the approximate nature of some of the elements in the estimates, this is a remarkable agreement between the two methods of computing the size of the spawning stock, and strengthens the view that the total estimate of eggs is siifficieuth reliable to warrant the conclusion that the egg production was in the order of 50,000 billion in 1932. This, of course, refers only to the spawning in the region south of Cape Cud, and it has been pointed out (p. 160) that important spawning occurs also in the Gulf of St. Lawrence. Since spawning in the latter region seemed to be of lesser magnitude than south of Cape Cod, it is probable that the entire spawning oil' the east coast of North America would not be more than double the estimated 64,000 billion, or, since the latter is an uncertain figure, let us say in the order of one hundred thousand billion eggs. SPAWNING HABITS

According to Bigelow and Welsh (1925, p. 208), "Mackerel spawn chielly at night." If this be true, the earliest egg stages should be relatively more abundant at certain times of the day than at others. From material collected at a number of stations in 1929, the eggs in "early cleavage" and "late cleavage" were counted, representing respectively the first and second 10 hours of development at the temperatures prevailing at the time. If spawning took place chiefly at night the early cleavage eggs should predominate between midnight and 10 a. m. and be in the minority during the remainder of the day. At the 14 stations from each of which more than 10 eggs of both stages were examined, the average percentage of early cleavage in the midnight to 10 a. in. group was 45 and in the 10 a. m. to midnight group 33. The difference between the two groups was not statistically significant(f=0.91 and P=0.3 +, according to the method of fisher, 1932, p. 114) and it may be concluded that the diurnal variation in percentage of early stage eggs does not indicate a tendency toward more spawning by night than by day. Tabulation of percentages according to the hours of the day did not indicate that any other particular part of the day was favored. 166 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE THE EGG

— 4 for Description.- According to published descriptions, (Ehrenbaum, 1921, p. for the European mackerel ;Dannevig, 1919, p. 11, and Bigelow and Welsh, 1925, p. 208, the American mackerel) the mackerel egg is 0.97 to 1.38 mm. in diameter and contains of taken at sea an oil globule 0.28 to 0.35 mm. in diameter. Measurements eggs far the commonest during this investigation had a similar range in dimensions. By dimension (modal) was 1.2 to 1.3 mm. for the egg and 0.31 to 0.32 mm. for the oil globule. ad- There is a tendency toward a decrease in size of mackerel eggs as the season vances. Data given by Ehrenbaum (1921, p. 4) show the same tendency in the egg of the European mackerel. This could be due to the seasonal trends of either tem- of Fish who found cod perature or salinity, but the experiments (1928, pp. 291-292), fertilized in warm eggs fertilized in cold water to be larger than those water, suggest that temperature alone could be responsible. Whatever its mechanism, the phenom- enon of decrease in size as the season advances probably holds true for all species to make scatter occuring in the tows of the present investigation. It was my practice diagrams in which oil globule diameter was plotted against egg diameters for all eggs in hauls containing troublesome mixtures. Invariably, when mackerel eggs were near the limits of the over-all range of their dimensions and thus might be expected the cor- to overlap the range of the eggs of other species, the latter were also near responding limits of theh respective over-all range and the groups remained discrete, showing that tendencies for smaller or larger than average size were shared simul- in dimensions was taneously by all species. Thus, in individual collections the range much less than the relatively large range of all collections, and a feature that might have been a hindrance in identification was in reality not very troublesome. In the collections made during the course of this investigation there were eggs of of the four species whose dimensions approached those of the mackerel. The egg common bonito (Sarda sarda) is 1.15 to 1.33 mm. in diameter, but in its early stages it has a cluster of small oil droplets instead of a single large one. In its late stages, there these droplets often become united into a single oil globule. In this condition it that bonito occur might be some difficulty in distinguishing the two, were not eggs consid- later in the season (hi areas we have prospected) when the mackerel eggs are erably smaller. For instance: Mackerel eggs taken in Cape Cod Bay, July 19, 1929, were 1.00 to 1.12 mm. in diameter while bonito eggs taken July 25, 1929, in the offing of No Man's Land were 1 .12 to 1 .27 mm. in diameter. The eggs of the cusk (Brosmius brosme) and the tilefish (Lopholatilus chamaehonticeps) are similar in size but have oil globules distinctly smaller (0.19 to 0.23 mm.) than those in the mackerel's eggs. Closer to the mackerel egg hi its dimensions was that of a species not yet identified. Although overlapping the mackerel egg in dimensions, its modal size was distinctly smaller and the oil globule somewhat larger, and in its late stages the embryonic pigment was arranged in bars unlike the diffuse arrangement in the embryo of the mackerel. Inasmuch as eggs of this type were found only at the edge of the con- and tinental shelf, their distribution was discontinuous with that of the mackerel; since no mackerel larvae were later found in the same or neighboring localities this egg caused no confusion. — in Bate oj embryonic development. Although mackerel have never been observed the act of spawning, it is generally supposed that both eggs and sperm are discharged into the surrounding water, where fertilization takes place. Observations have shown BIOLOGY OF THE ATLANTIC MACKEREL 167

6 that thereafter, during the period of embryonic development, the eggs are suspended in the sea water mostly near the surface and all above the thermocline. As is true with most cold-blooded organisms the rate of development depends on the temperature at which it takes place, being slower at low temperatures and faster at high temperatures. According to Worley (1933), who examined this feature of the development at the U. S. Fisheries Biological Station, Woods Hole, Mass., the time elapsing between fertilization and hatching was 50 hours at 21°, 70 hours at 18°, 95 hours at 16°, 115 hours at 14°, 150 hours at 12°, and 208 hours at 10°. There is no reason for believing that the rates differ at sea, though this is difficult to demonstrate. According to Worley (1933, p. 857), "Experiment showed that typical development (and survival) could be realized only between 11° and 21°." At sea in 1932, however, eggs were most abundant at temperatures below 11°, as appears from the following average numbers taken at each degree (centigrade) of surface temperature encountered in the survey:

7 - 14 150 8 HI 15 555 9 2, 117 16 44 10... - 3,360 17 5 U 2,432 18 74 12 - 1,390 19 13 1,380 20.__-

The embryos in eggs from water below 11° C. differed in no perceptible way from those found in warmer water, and there is no reason for believing that development was not proceeding as "normally" at the lower as at the high temperatures. the incubation Worley also found (loc. cit.) that "The total mortality during period was least at 16° C. where it amounted to 43 percent." He had three experiments at this temperature with mortalities of 37, 40, and 53 percent respectively inter- (loc. cit. p. 847). At sea, in 1932, the average mortality was 59 percent (from or a polation to the hatching point from the data of the 5th column in table 7), only little greater than in the least favorable of the laboratory experiments. The weighted mean temperature of the water from which these sea-caught eggs were taken was 10.9° C. Worley's laboratory eggs suffered 90 and 95 percent mortality in his two experiments at 11°. Obviously, both the range for normal development and the point of maximum survival were at lower temperatures at sea than in the laboratory experiments of Worley. The explanation for this disparity between results in the laboratory and observations at sea probably lies in the fact thai Worley's experiments took place at a time when temperatures of the sea water from which he took his fish were in the neighborhood of 10° C. The lesser mortality at and near this temperature was connected no doubt with the lesser change involved in bringing the eggs from the temperature of the parent to the temperature of the experiment. It is obviously de- sirable that laboratory experiments be repeated on material taken from water of lower temperature. Vertical distribution.—Although it has been known that mackerel eggs are sus- waters no pended in the sea, usually near the surface, there has been in American previous determination of vertical distribution, apart from the general observation

« For the minutiae of the embryology of mackerel, the reader is referred to Moore (1S95, pp. 5-14), and to Wilson's (1891) desr-rip tion of the sea bass, which the mackerel in its embryology closely resembles. 168 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

that surface hauls take more eggs than deeper hauls. The present determination is based on a series of horizontal hauls at different depths in 38 meters of water in the offing of the Fire Island Lightship on May 19, 1929. Four series were taken: one at dawn, another at noon, another in the evening, and the final series at midnight. The net was one-half meter in diameter at the mouth and rigged with a closing device actuated by a messenger. It was lowered while open, towed for 20 minutes, then closed and hauled to the surface. Each series the 10- included hauls at the surface and at 5-, , 20-, and 35-meter depths. The courses of the nets were kept as nearly horizontal as possible by periodical estimation of depth based on measuring the towing warp's angle of stray and paying out or hauling in the line as needed to keep the net at the proper level. Since the net was lowered while open, and since the tripping mechanism failed on several occasions, there was some contamination of the haul during its passage through the water over- hang the stratum fished. Correction for this contamination was estimated on the basis of the average concentration of eggs in the overlying water and the time it took the net to pass through the overlying water in an opened condition. An additional correction for variations in speed of towing, based on the angle of stray of the towing warp, was applied to all catches on which data adequate for this purpose were available.

Table 2.— Vertical distribution BIOLOGY OF THE ATLANTIC MACKEREL 169 took very few eggs that were not otherwise accounted for (by the contamination correction based on the average catch of the upper net and on the time taken to pass through the upper stratum). It is safe to conclude therefore, that the pycnocline forms a barrier to the downward extension of mackerel eggs. Further, the pycnocline latter be is sufficiently well indicated by the thermocline in this region so that the may used an an indicator of the lower limit of mackerel eggs.

N'JMO-R OF THOUSANDS OF MACKEREL EOOS PER HAUL TEMPERATURE, DEGP.FES C. 5 10 !5 in 30 « 8 1? 12 SURFACF

BOTTOM 335 3-10 3Z3 3?-0 31.5 SALINITY, PARTS PER THOUSAND

t.C'255 I.C250 '.0245 Of.NSITY IN SITU

ii it of water. Observations were Figure 4.—Vertical distribution of mackerel eggs in relation to temperatures, sa n y , and density adjusted to the basis of standard speed of towing, except tbose indicated as questionable.

differences in the The serial tows of May 17, 1929, also illustrated significant vertical distribution of eggs in different stages of embryonic development. When the eggs were separated into three stages of development occupying approximately equal the were near the periods of tune, it was found that those of early stage (A) mostly 5- 10-meter and surface, those of the late stage (C) mostly between the and levels, those of the intermediate stage (B) intermediate between A and C in their vertical to indicate distribution (table 3). Too few eggs were taken at greater depths reliably the proportionate numbers at each of the three different stages of development.

station Table 3. — Vertical distribution of various stages of mackerel eggs according to noon series, 20498, May 17, 1929

is from tn extending three quarters around tho [Stage A is fmin fertilization to complete epiboly; stage B complete eplboly embryo circumference of the egg; stage C is from this point to hatching] 170 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

7 his 3-day-old eggs nor that of the sea water either at the beginning or end of his experiment. Since he was working before rigid control of temperature was cus- tomary, it is likely that the specific gravity of the water in his experiment may have been changed by warming. In the present example, at least, it is known that the temperature of the water was increasing at the time station 20498 was visited. At the temperature of the water in which the eggs were found on May 17, it takes about 5 days for incubation (p. 167), and it may be estimated that stage C eggs were spawned at least 3 days prior to stage A eggs, hence on May 14, when unfortunately this station was not visited. However, from interpolation (linear) both in space and time between the temperature at station 20498 on May 17 and temperature at neighboring stations on May 12, it appears that the density of the water at the surface on May 14 could have been very nearly the density of the water at the 10-meter level on May 17. Hence it is preferable to ascribe the sinking of the late stages to the warming of the water with attendant decrease in density, rather than to an increase in the density of the eggs.

THE LARVA '

9 Yolk-sac stage.— The newly hatched larva is slightly less than 3 mm. in length, well covered with scattered black pigment spots which tend to be denser dorsally than vcntrally. The eyes are colorless. The region of the gut is occupied by the yolk sac with its oil globule. Both sac and globule are about the same size as they were in the egg. The mackerel is readily distinguished from other similarly marked larvae with which it is found, by its larger size, stouter shape, coarser pigment spots, and its 30 myomeres. As development proceeds, the pigment becomes localized on top of the head and along dorsal and ventral edges of the body, the eye becomes black, the yolk sac absorbed, the mouth and gut formed. These changes are completed at a length of 4 mm. As seen in the laboratory and hatchery, the mackerel swim very feebly during the yolk-sac stage, with short, spasmodic, random movements. Their balancing faculty is undeveloped, their position being indifferently upside down, right side up, and at various angles. At sea they must be totally at the mercy of the water movements. — Larval stage. As used herein, this stage represents the period beginning after yolk-sac absorption and ending after fin formation, and it includes individuals between 4 and 8 mm. in length. In this stage, the mackerel is readily distinguished from other species by the row of black spots of irregular size and spacing along dorsal and ventral edges of the body, beginning about midway between snout and tail and extending almost to the end of the notochord (but not into the fin fold). Those in the dorsal row are less numerous and more widely spaced than those in the ventral. Other species which were found with the mackerel, and which have also such dorsal and ventral rows of are the winter flounder pigment, (Pseudopleuronectes americanus) , which differs from the mackerel by its greater number of myomeres (37-40) and its

' But he does give the specific gravity of newly spawned, eggs as between 1.02-1 and 1.025, a figure very close to that of surface water at our station 20498. (See fig. 4.) > While, the term larva may be applied to the entire planktonic existence, it is convenient to recognize three subdivisions: yolk- sac stage, larval stage, and post-larval stage. • This description is based on formaldehyde preserved specimens because this is the form commonly available for study. In life, the newly-hatched larva is longer, measuring 3.1 or 3.2 mm. (distortion and shrinkage decrease the length of preserved specimens), and in addition to the black pigmentation, have yellow and greenish pigment on each side of the head between the eye and otocyst, am! on the surface of the oil globule (Ehrcnbaum, 100S p. 31). BIOLOGY OF THE ATLANTIC MACKEREL 171 strongly, laterally compressed body; the bluefish (Pomatomus saltatrii), which differs its fewer and the rosefish which has the same by myomeres (24) ; (Sebastes marinus) , number of myomeres (30) and in the 4- to 5-mra. stage could be confused with mackerel. With both the rosefish and mackerel available for comparison, the former is readily distinguished by the closeness of the spots in the dorsal and ventral rows, those in the rosefish forming almost a continuous black streak, whereas those of the mackerel are discrete. Other differences, less useful, are the more slender shape and the greater relative length of the post-anal region in the rosefish larva. After passing the 5-mm. stage, the rosefish larva is readily separated from the mackerel larva by its prominent preopercular and cranial spines. An additional character of use in separating the mackerel larva from the others is its strong teeth, which are readily visible in specimens of the 7-mm. size but less so in smaller individuals. Inability to keep larvae alive in the laboratory or hatchery during this stage precluded direct observation on their activity, but, as is shown in a later section, their movements are sufficiently well-directed for performance of diurnal vertical migrations of 20 to 30 meters but not sufficiently sustainable for migrations of miles in extent. Transition phase. —Intervening between larval and post-larval stages is a transition phase including individuals 9 and 10 mm. long whose fins are in various states of 10 completion. Fin formation is a gradual process, neither beginning sharply at 9 mm. nor ending sharply at 10 mm. At the former length, the caudal fin already shows a number of rays, and at the latter length, the laggard first dorsal fin does not 3"et show any of its spines. But the tail fin makes its greatest changes, the second dorsal fin and finlets and the anal fin and finlets are all developed within this size range, hence it is most appropriately designated as a transition phase. Post-larml stage. —This stage includes the latter part of planktonic existence beginning at about completion of fin formation and lasting until the young (ish arc nimble enough to evade the plankton nets. It is comprised of individuals 11 to 50 mm. long. Since all the vertical fins except the first dorsal are complete, identification by adult characters is simple. The larvae enter this stage somewhat laterally compressed, and by its end fill out to the trim fusiform shape of the adult. At the beginning of this stage the color pattern is typically larval, but by its end the dark pigment has spread over the dorsal portions, and in live specimens the silvery line is apparent, though the black wavy bands characteristic of the adult are yet to form. The appearance is in general like a miniature adult with somewhat oversized head and fins. As appears in a later section, the post-larvae are capable of extensive swimming. Furthermore, as they near the end of this stage the schooling instinct asserts itself. The transition from a primarily planktonic habit to a primarily swimming and schooling habit probably is gradual, in the sense that all individuals may not experience the change at the same size. The available evidence is that it involves individuals between about 30 and 50 mm. in length. This evidence is from two sources.

First, the survival curve (fig. 17) has a substantially uniform trend from 11 to 30 mm., from which it may be inferred that there was no change of trend within this size range sufficient to indicate a loss of larvae such as could be expected if some had begun to

"The present description ot lengths at which Cns appear differs from published figures (Ehrenbaum, 1921, figs. 1 to 7, and

Bigelow and Welsh, 1925, fig. 92) probably because the latter give lengths inclusive of finfold or caudal fin, though this is not definitely stated; whereas our measurements were taken to the end of the notochord, i. e., exclusive of the finfold in early stages; and to the base of the caudal fin rays, i. e., exclusive of the caudal fin in later sts'ees. This was necessary on account of fre uienl ''' tortion or injury to the caudal appendage. 172 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE school and were no longer susceptible to capture in plankton nets. Second, a school of small mackerel was observed and sampled in Woods Hole Harbor in July 1926, containing individuals between 35 and 65 mm. in length (table 21). The first evidence shows that the schooling habit did not involve fish under 30 mm. in length; the second proves that some fish, at least, begin schooling as soon as they exceed that size.

Vertical distribution.—From series of horizontal hauls at 0, 5, 10, 20, and 35 meters at early morning, midday, evening, and midnight, at a station (Albatross II No. 20552) southeast of Fire Island Lightship (latitude 40°20' N., longitude 70°57' W.) visited on July 13 and 14, 1929, there is evidence that the larvae of the mackerel do not descend far below the surface, probably being limited by the thermocline, and that they perform a diumal vertical migration (fig. 5).

Table 4.— Vertical distribution of mackerel larvae] at various limes of the day as indicated by horizontal lows with a closing half-meter plankton net at Station 20662 (Albatross II), latitude 40°20' N., longitude 72"69' TV., July IS and 14, 1929

Depth of haul BIOLOGY OF THE ATLANTIC MACKEREL 173

It is improbable that the daytime descent was beyond the 20 meter level at this station or was ever beyond the thermocline. During 1930, 1931, and 1932, when the nets were hauled obliquely below as well as above the thermocline, the lower tows seldom caught larvae that could not be accounted for as contaminants resulting from passage through the upper layers. From the length-distribution of the larvae it appears (table 4) that the larger individuals (6 to 9 mm.) were more stongly inclined to migrate, reaching the surface at night, while the smaller ones (4 to 5 mm.) tended to stay in the intermediate 5- to 10-meter levels. Though these observations do not provide a precise description of vertical distribution and migration, they do demonstrate the necessity of sampling all levels down to the thermocline to get the representative statistics needed for the studies on growth and mortality to follow.

2E -i 1 r-

120 t

MIONIGHT Dawn S 10 12 14 18 18 T gMPER«TU3r., *c.

Figure 5.—Vertical distribution of mackerel larvae at several points of time in the dirunal cycle in relation to temperature. Tn 6 solid lines connect observational points. The broken lines indicate the probable vertical position of the bulk of the population of larvae. GROWTH

Very little has been published on the growth of marine fishes during that early period of the life history spent in the plankton community, and nothing on the growth of the mackerel during this stage. Of the data collected during the present investigation, only those of 1932 were collected in a manner sufficiently quantitative and at short enough intervals of time to be used in deducing growth rate. The method of analysis consists, essentially, in following the advance in position of the mode of homologous groups of larvae by comparing sizes collected in successive cruises. But this cannot be done in a simple and direct manner. Mackerel eggs are spawned over a period of several months. The larvae are subject to high mor- small larvae than tality. As a result, almost always there are vastly more large ones, and the predominance of small larvae is so great during most of the season that the arithmetic groups of larger ones do not form distinct modes. Instead, in ordinary frequency distributions they are apparent principally as a lengthening of tho "tail" of the distribution at its right-hand side (table 5). 174 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 5.—Number of eggs and larvae taken on each cruise in 1932, classified according to stages of eggs and le7iglhs of larvae

(During cruises 1 to 7, tow nets 1 meter in diameter at mouth were used, and duriog cruises 8 and 9, tow nets 2 meters in diameter were used; all hauls were obliquely towed and numbers caught were adjusted to represent an equal amount of towing per meter of depth fished] BIOLOGY OF THE ATLANTIC MACKEREL 175

That each scries includes truly homologous groups is indicated by several criteria, independent of the straight-line conformity. In the R series, the modes all tend toward peakedness. In the S series, they all tend to be broad. In the T series they are intermediate in shape. The progress in each series is reasonably consistent and the course of growth is roughly parallel in the three series; moreover, the slight depar- ture from parallelism is in the expected direction, the later series having the higher growth rates consistent with their development in the warmer water to which they are subjected. Furthermore, the modes are consistently present in the material from each cruise with only two exceptions, R in cruise III and S in cruise IV. The absence of S in cruise IV is plainly due to failure on that cruise to visit certain stations in the southerly end of the spawning area, where previous cruises woidd lead one to expect

4.0 176 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 6.— Deviations of individual cruise frequencies of lengths of larvae and postlarvae from the average frequency ' of the 9 cruises of the season of 1932 BIOLOGY OF THE ATLANTIC MACKEREL 177

11 and 1927 when several samples were secured by dip net early in summer. Their lengths (table 21) agree closely with the terminal position of the growth curves described by the chosen homologies, and are far below a growth curve predicated on the alternatives. Hence it may be concluded that the chosen series consist of truly

0.6 0.8 1.0 1.2 1.4 1.6 LOG OF LENGTH, MM.

_1 L_1_L 10 15 20 30 40 50 LENGTH, MM.

Figure 7. —Growth or mackerel larvae and post-larvae as indicated by the progress of modes in the deviations of numbers of specimens in each size-class taken on individual cruises from the average number taken on all cruises. The letters R, S, and T mark the positions of homologous modes referred to the scale of dates; and the straight lines are fitted to the homologous series. The vertical interior scale is the scale of deviations in logarithms. Roman numerals are cruise numbers. homologous modes, and that the straight lines fitted to the respective series correctly describe the larval and post-larval growth in 1932.

» Schools of very small mackerel wander into pound-nets from which they can be removed by dip net if the pound-nets are visited before hauling. Once hauling commences they are frightened and usually escape through the meshes. In addition tu samples so collected, [one was taken from a school which wandered into the boat basin at the U. 8. Fisheries Biological Station, Woods Hole.

[library 178 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

2 2 BIOLOGY OF THE ATLANTIC MACKEREL 179

incubation period, stage B, from complete epiboly to embryo % around the yolk mass constitutes 32 percent, and stage C from embryo % around the yolk mass to hatching constitutes 33 percent (Worley 1933, fig. 5). The average time occupied by these three egg stages was therefore 2.53, 2.31, and 2.39 days, respectively, and the average age of each stage was derived by simple aritlimetic. The duration of each larval length-class was computed from the formula:

duration ' (in days)= °o 01591

where li is the lower boundary of the length class interval in mm., 1 2 the upper boundary of the length class interval in mm. The constant 0.01591 is the increase per day of the logarithm of lengths computed from the straight line fitted to the

points of tbe S series (fig. 8). The average age of each length-class was computed by the formula:

1 age (in days)= -^=^f^+7.23

where h is the length of newly hatched larvae (2.8 mm.) and 1* the midvalue of the length class interval. Tbe constant 7.23 is the average age of newly hatched larvae. The boundaries of class intervals were as follows: for 3-mm. larvae, 2.9 to 3.5 mm.; for 4- to 25-mm. larvae, the designated length ±0.5 mm.; for 30- to 50-mm. larvae, the designated length ± 5.0 mm. The mid values of class intervals were: for 3-mm. larvae, 3.2 mm.; for all others, the designated lengths. Accuracy oj determination.—The resulting values for duration of egg stages and of larval-length classes are given in table 7 to hundredths of days, thus expressing a smooth curve that gives the most probable relationship for the body of data from which they are derived. Purely from the standpoint of instrumental and sampling accuracy, they have no such high degree of precision. The durations may be accurate to the nearest tenth of a day for the egg stages, and of lesser accuracy for the larval-length classes. The duration of the 3-mm. class, derived by extrapolation, is especially in doubt, and may be in error by as much as a day. The other classes probably are within several tenths of a day of true values. From the standpoint of variability in growth itself, the values are even more approximate. While growth obviously follows a curve of percental increase, there must be fluctuations about this curve due to local variations in environment affecting accessibility of food and rates of metabolism. Furthermore, the particular curve of growth given pertains only to the S group, which developed under a particular set of environmental conditions. From figure 8 it appears that the earlier hatching R group, developing, on the whole, in cooler water, grew more slowly than the S group, while the later hatching T group grew faster in the generally warmer water in which it developed. Thus the R group took 56 days, the S group 50 days, and the T group 47 days in growing from a length of 4 to a length of 25 mm., a divergence from the S group of 12 percent in one instance, and 6 percent in the other. This is by no means the extreme variation to be anticipated, for it is conceivable that temperature or other influences might vary more widely than happened in these three instances, and correspondingly greater differences of growth would follow. On the other hand, the S group developed from eggs spawned somewhat early in a season that was slightly warmer than average (Bigelow, 1933, p. 46) and thus in temperatures that would likely be reproduced in the middle portion of less unusual seasons, and therefore 525293—44 3 180 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE the rates computed from the S group must be near the usual rate, probably within 10 percent. Discussion of growth. —Having determined the rate of growth of the mackerel through its early life, it would be interesting to have comparisons of the early growth of other fishes, particularly to see if logarithmic growth is the general rule. Unfor- tunately, there is a paucity of data on this subject, most of the material on growth of fishes being confined to the portion of life following the larval or post-larval stages. From various sources, however, it has been possible to assemble material on the early growth of three other species: the herring (Clupea harengus) in the Clyde Sea area, the haddock (Melanograrnmus aeglifinus) in the waters off the northeast coast of the United States, and the northern pike (Esox lucius) of North American fresh waters.

1.2 -

1.0

• 8 < 40 BIOLOGY OF THE ATLANTIC MACKEREL 181

9, it is seen that logarithmic curves with a change in slope at 30 days of age, or length of 19.5 mm., fit the points as well or better than does the straight line in the lower part of figure 9. The observations on haddock (Walford, 1938, p. 68-69) were taken in a manner similar to those on mackerel. In fact, the material consisted mainly of haddock larvae caught on our mackerel cruises. Walford summarized these by months, giving frequency distributions for each of the four months: April, May, June, and July. From these polymodal frequency distributions, he selected modes that he considered to be homologous, recognizing three such series. Taking his middle series as perhaps the most typical, the modal values, as nearly as can be read from his figure 50, were

50 40 30

20 i

10 5

6 ->

- - 4

0.4 10 20 30 40 50 60 70 60 90 100 AGE, DAYS

Fioori 10.—Growth of haddock during early lite. Data from Walford, 1938.

3, 3.5, 18, and 43 mm. on the mid-dates, April 11, May 15, June 17, and July 17, respectively. According to Walford, the 3 mm. mode of the first cruise consisted of recently hatched individuals. Assuming this size to be zero days old, the logarithms of the modal sizes were plotted against age in figure 10, whence it is apparent that the 12 growth of the haddock was logarithmic as in the mackerel. Data on the northern pike (Embody, 1910) consisted of the average length in samples of two or more specimens drawn from a population reared in the laboratory at water temperatures of 65° to 72° F. Since the data are not readily accessible, they 13 are repeated below :

Total length in . Total length in in after Age iu days after hatching: \miltimcters Age days hatching: mllllmeteu 7 5 13

2 9. 25 7 '14 3 10. 5 9 15.25 4 11.5 11 16

! Sac absorbed.

» Another of the series of modes selected by Walford also becomes logarithmic with slight re-interpretation of his fig. 49. The new interpretation involves the assumption that the group in question was under-represented in the April sample, an assumption that is reasonable in view of the fact that his samples for this month were from a more easterly area than that subsequently sampled. (This is true also of the central mode, above discussed, but the group forming this mode could have drifted into the area subsequently sampled, whereas the time sequences were such that the group here under consideration in all probability could not have so drifted) . It further involves taking the mode for May at 12 instead of 17 mm. and for June at 30 instead oi 33 mm. These selections are 01 prominences on the curve, which are equal to those selected by Walford, and by reason of parallelism with the middle group, seem more reasonable than the points given in Walford's figures 49 and 60. Walford's third series obviously consists of a younger group not present enough months to repay study. 11 1 am grateful to the late Professor Embody for communicating these data to me by letter. 182 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Plotted on a logarithmic scale, these values describe the curve given in figure 11. It is interesting to note that the change in slope approximately coincides with yolk sac absorption. For ready comparison the growth curves of mackerel and of these other species are assembled in figure 12. In all of them, length was used as an index of size. Mass or volume would be a more nearly true index. However, if there is no change in form, length would serve well to test for logarithmic growth since a certain power of length would be proportional to the mass or volume, and in logarithmic plots the only difference between the two would be a difference in vertical scale. Since the mackerel and haddock undergo little change in form during early life history, a simple logarithmic curve well fits their growth as indicated by length. The herring larva, on the other hand, is slender and almost eel- like when young, growing stouter as de velopment proceeds. This being true, length overestimates size early and under-

- 20

15 z x 10 |

5 l( AGE, DAYS

Fiouee 11.—Growth of northern pike during early life. Data from Embody, 1910. estimates it later. This may be the reason for the nearly linear arrangement of points when lengths are plotted directly against age. Further, the change in slope when the logarithms of lengths are plotted against age suggests that the change in form is greatest at about 30 days of age when the herring is about 18 mm. long. The growth of the northern pike, too, shows a change in slope. In this instance it approximately coincides with yolk sac absoption, hence this might as easily be a real change in growth rate due to difference in food availability or assimilation rather than an apparent change due to altered form. Evidence from the information available on these several species supports the view that growth in the early life of other fishes, as well as the mackerel, is logarithmic in character and at a uniform percental rate throughout this stage of life except when there is a change in mode of living (e. g., yolk sac absorption) and that the use of length as an index of size may complicate interpretation of growth rates when there is considerable change in form. BIOLOGY OF THE ATLANTIC MACKEREL 183 DRIFT AND MIGRATION

The current system in the waters overlying the continental shelf between Cape Cod and Cape Hatteras has yet to be studied. Evidences collected during this in- the vestigation from releases of drift-bottles and computations of dynamic gradients, latter subject to large errors of interpolation, were not sufficiently conclusive to deserve publication. They indicated slight tendency for movement in a southwesterly direction parallel to the coast, probably not strong enough to transport eggs and larvae of the mackerel important distances. On the other hand, evidence from the distribution of mackerel eggs and larvae themselves leads to definite conclusions. From the growth curve of larvae, figure 8, or from the position of homologous modes in the deviation curves, figure 7, it is possible 50 40

8 uj

0.4 10 20 30 40 50 60 70 80 90 100 AGE, DAYS

Figure 12.— Growth of northern pike, herring, mackerel, and haddock. to ascertain the lengths attained by certain groups of larvae at each successive cruise. By plotting the geographical distribution of larvae of these particular lengths in successive cruises, as in figure 13 based on the S series, their movements may be followed. In general, this series represents a population spawned over the continental shelf off the New Jersey coast. Larvae hatched from these eggs remained in this area until they reached a length of 8 mm. about a month later. Thereafter, there was a northeasterly shift which brought the population to the region just south of Long Island at the end of their second month when they were about 20 mm. long. Movement toward the northeast probably persisted still longer, for the only individuals the east large enough to have been members of this series were taken at stations along coast of Massachusetts (Chatham II and Cape Anne II in table 20) during the cruise of July 14 to 28. Although there is local spawning in Massachusetts Bay, it is unlikely that it was responsible for these large individuals, because spawning usually is later as in Massachusetts Bay, and the locally produced larvae could not have grown to large a size as the 37- and 51-mm. post-larvae taken on July 22. 184 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Examining in greater detail the distribution in the successive cruises, two concentrations were evident within the area over which the larvae of this series were distributed. One may be called the northern center; the other, the southern center. u The northern center was off the northern part of New Jersey (New York II) in the

76* 74* 72' 70° BIOLOGY OF THE ATLANTIC MACKEREL 185

The southern center shifted southward from off Delaware Bay (Cape May II) half way to the Chesapeake Capes (Winterquarter I) where it remained during the following cruise and possibly the next one also, though these stations were not visited on the fourth cruise. During the fifth cruise it was found farther north and seaward in the offing of the southern New Jersey coast (Atlantic City III and Cape May IV). Next it appeared to join the northern center and was apparent as a tongue extending from this center to the offing of the middle of the New Jersey coast (Atlantic City II) . Thereafter its location apparently coincided with the northern center. During the time that the two centers were separate they moved in essentially to 22 and then identical directions (fig. 15). Both moved southward from May 3 May of northward until June 7, apparently under a common impulse. If the resultants 16 wind direction and force during the cruises be plotted, as in figure 15, it is seen that the strong winds blew in essentially the same direction as the larvae moved, southerly until May 22 and then northerly until June 7. Obviously the wind, by drifting the surface water, was responsible for the transport of the larvae. After June 7, however, the movements of larvae did not correspond so closely with the movements of the wind extent of them. Thus the move- (fig. 16) and must have been to some independent ments of the population of mackerel larvae may be divided into two phases, an early passive phase and a later active phase. The break between the two came, as might be expected, when the larvae, at a length of 8-10 mm., developed fins (p. 171) and graduated from the larval state to the post-larval stage. The movements in the two stages will be considered in detail separately. During the passive phase, although the movements of the two centers of larvae are essentially similar and both correspond to that of the wind, there are minor differences worthy of note. The southern center was found at the same place on cruises II and III in spite of considerable sustained wind from the northeast and corresponding movement of the northern center in the interim between the cruises. Later there was the great shift of the southern center between cruises III and V without correspond- drift of the northern ingly great wind movement and without correspondingly great center. To some extent these discrepancies may be due to failure precisely to locate 16 centers of distribution with the stations as far apart as of necessity they were. But it is more likely that the peculiarity in the relation of the drifts of the northern and southern centers has a physical rather than statistical basis. The outstanding in direction almost peculiarity was that the northern center traced a course a southerly whereas the equal in distance to its return in a northerly direction (up to cruise VI) southern center moved southerly a much shorter distance and then returned northerly it is noticeable a much greater distance. Considering now the topographical features, that at the northern and middle portions of the area the continental shelf is broad and the water relatively deep, while at its southern end the shelf narrows sharply and the water is much shoaler. A water mass impelled by the wind could move in a southerly either: direction freely until it reached the narrow, shoal southerly end where it must the southern turn (1) stream very rapidly through the "bottle neck" at extremity; (2) out to sea; or (3) pile up temporarily.

or " Records of the Winterquarter Lightship, 8 a. m. and 4 p. m., including only those winds of force 3 (Beaufort Scale) higher, were plotted in vector diagrams to determine the resultants. the chart of movement » The true position of the northern center at the time of cruise III (fig. 13) was particularly uncertain. On northern stations with catches, that is, Atlantic (flg. 15) it seemed logical to plot it at the center of gravity between the three largest there or elsewhere, and hence missed. City II, Cape May I, and Cape May III. but its true position most likely was between stations, This accounts also for the almost complete obliteration of mode S on this cruise, to which attention was earlier called in discussing progress of modes as indicating growth. 186 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

75" 74° 73* 72" 71" 70*

Fiouke 14.—Location of stations occupied during the 9 cruises of 1932. BIOLOGY OP THE ATLANTIC MACKEREL 187

That it did not do (1) or (2) is proved by the relative scarcity of larvae of appropriate sizes at stations of the Chesapeake section and the outer station of the Winter- quarter section; though the few caught at Chesapeake II, III, and Winterquarter III indicate a slight tendency for southward and outward streaming. That (3) was the major result is shown by the "snubbing" of the southern center in its southward travel and by the increase in numbers of larvae in the southern center relative to the number

Figure 15—Drift of the 2 centers of distribution of the S group compared with wind movements, as recorded at Winterquarter Lightship.

17 in the northern center, as if indeed the water and its burden of larvae did pile up in the vicinity of Winterquarter I. This piling up very likely was in the nature of a thickening of the surface stratum of light water offset by a depression of the lower layers of heavier water rather than an outright raising of the water level. Of course, the depressing of the subsurface stratum would set up a subsurface flow to restore equilibrium. This flow would not transport the main body of larvae, since they were

17 It is not supposed that the entire increase in relative number at the southern center was due to the mechanism being discussed. Part of it could have been due to random fluctuations of sampling. 188 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE confined to the upper stratum (p. 172); it could and probably did carry a few that happened to be near the interface as indicated by the light spread of larvae southward and outward to Chesapeake II and III and Winterquarter III. While this accounts for the halting of southerly drift of the southern center and its increase in relative numbers, there is still to be considered the apparently too rapid drift of this center northward when the wind direction was reversed. Let it be supposed that the aforesaid piling-up of surface waters took place more rapidly than could be counterbalanced by subsurface flow. Then the sea surface would actually have risen and remained at a higher level as long as the wind continued to transport surface water to the area faster than the subsurface water could flow away. Then when the wind reversed its direction, the energy so stored would be released and act in the same direction as the wind. The two forces together would produce a faster drift than could result from the wind force alone, and thus account for the high rate of movement of the southern center between May 22 and June 3. Whether the interactions of the wind forces and water movements here postulated were theoretically probable from dynamic considerations must be left to the physical oceanographer. He can find here an example of biologically marked water probably of considerable aid in the deciphering of the pattern of circulation in shallow water, where difficulties of dynamic analysis are heightened by topographical features, and where a better understanding would be of greatest practical use in dealing with fishery problems. Whatever the outcome of any future examination of the dynamics of this situa- tion, the outstanding resemblance of the main features of wind movement to larval drift, together with the fact that deviations from the parallelism between the two have a plausible though not proved explanation, leaves no doubt that the larvae (and the water with which they were surrounded) were drifted from place to place by the wind's action on the water, and that this alone accounted for their movements until they reached the end of the larval stage at a length of about 8 to 10 mm. and entered upon the post-larval stage. Subsequently the movement of larval concentrations corresponded less perfectly with that of the wind (fig. 16). Between cruises VI and VII, when there was a gentle easterly wind movement, the post-larvae also moved eastward, but proportionately father than might have been anticipated from the moderate wind movement. Between cruises VII and VIII, when there was a northeasterly wind movement, they moved northwesterly. After cruise VIII it is difficult to be sure of the homology of the group under consideration, but the only post-larvae (lengths 37 and 51 mm.) of cruise IX identifiable as belonging to this group were caught at Chatham II and Cape Ann II, off eastern Massachusetts. The indicated movement was in the same general direction as the prevalent strong winds, but again sufficiently divergent to indicate some independence. Since the drift of water under impulse from the wind accounts for only a portion of their movement and since such evidence as is available on 18 residual surface flow in this region indicates water movement westerly, hence in a direction contrary to the movement of the post-larvae, the evidence does not favor the transport of the post-larvae as purely passive organisms, and it must be concluded that they moved to an important extent by their own efforts. This is in complete harmony with their developmental history. As larvae, without swimming organs other than the rather flaccid finfold, they drifted with the

'• Drift-bottles set out by Wm. C. Herrington (unpublished data) in connection with his haddock investigations in the spring of 1931 and 1932 drifted westward past Nantucket shoals, fetching up on beaches of southern New England and Long Island. BIOLOGY OF THE ATLANTIC MACKEREL 189 current; as post-larvae, with capable fins, they were able to swim and exercised this in of faculty. The change in locomotive ability coincided with change method transport. Thus far, attention has been focused on the main centers of larval concentration. It will have been noted in figure 13 that there were indications of a smaller body of larvae not included in the groups whose centers were followed. This body probably became separated from the southern center about May 23, when the center was at its extreme southerly position, and, as previously pointed out (p. 187), there was a spread to Chesapeake II and III and Winterquarter III, probably consisting of only

76 75 73 70 69 —r— —r~

(7-23) IX O

WIND MOVEMENT

~3X O * nVTT (6-6) (6'

76 75 74 73 72 71 70 69

Figube 16.—Drift of post-larvae of the S group compared with wind movement, as recorded at Nantucket Shoals Lightship.

those larvae that were at the interface between the accumulating surface water and the outward streaming subsurface layer (p. 187). Having been caught in this outward and perhaps somewhat northerly flow, their northward drift could start sooner and would take place farther offshore than the drift of the southern center itself. With this in mind, it is easy to account for the catches at Atlantic City IV on cruise IV and at Montauk II and No Man's Land II on cruise VI. That they did not appear on other cruises is not surprising, for their numbers were few (1,1, and 2 were caught at the respective stations above mentioned) and as the result of chance fluctuations in random sampling they could easily fail to appear in our hauls. 190 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

The average rate of movement of the S group larvae during the period from May 4 to June 6, while they were dependent for transport on wind-impelled drift, was 6 nautical miles per day. As nearly as may be estimated from data recorded on the Beaufort Scale, the net wind movement in the direction of the resultant (neglecting forces under Beaufort 3), was about 60 nautical miles per day. The movement of the in center of post-larval abundance between June 6 and July 1, accomplished part by swimming, averaged 3% nautical miles per day. If the movement of post-larvae between June 27 and July 24 may be taken as from off Shinnecock to off Chatham, the average rate during this period was 6 nautical miles per day. The movements of the R and T groups of larvae can be traced in the same manner as were those of the S group. The R group, beginning with cruise I, as 3 to 5 mm. larvae, moved southward from the Winterquarter section to the Chesapeake section. Like the S group, they remained at this southern extremity of the range through cruise III and also probably through cruise IV, though during the latter cruise there were not sufficient stations occupied in this area to prove this. On cruise V, however, they were found to have moved northward to Cape May, and on cruise VII were discovered off Shinnecock. At the beginning of this northerly movement, they were already 8 to 10 mm. long, and thus capable of swimming. With favoring winds during all but the last portion of this northerly trip, their movement was rapid, averaging 1 1 nautical miles per day. The T group could not be so readily followed, but in general its movements were with the wind in the larval stage and indifferent to the wind in post-larval stages. Between cruises III and VI, when the winds were from the southwest, it shifted in an easterly direction from the Shinnecock section to the Martha's Vineyard section. The correspondence between wind direction and this movement was not as perfect as that of the S group, formerly described. From cruise III to cruise IV, there appeared to be a spread in both easterly and westerly directions, and between IV and V, there was a contraction toward the center of the group off Montauk Point. These changes in distribution may be indicative of spurts of spawning rather than movements of the egg population, for they occurred during periods of egg development, and the stages chosen may not have been exactly the continuation of the original first seen as stage A eggs of cruise III. It probably suffices to note that when stage A, they were off Shinnecock, and by attainment of lengths of 4 to 5 mm. at cruise VI, with they were off Martha's Vineyard. Then between cruises VI and VTI, only larval remained a slight wind movement from the west, the zone of densest population at Martha's Vineyard, though fair numbers were as far west as Shinnecock. Between cruises VII and VIII, while the winds were from the southwest, the members of this group spread over the waters abreast of Long Island, extending from the New York to the Shinnecock section. During this interval they had grown into the post-larval stage, 10 to 12 mm., when swimming activity made their movements fairly independent of the wind. It may be concluded therefore, that the movement of eggs and larvae (up to 10 Hatteras mm. in length) in the southern spawning area between Cape Cod and Cape was governed by the drift of surface wateis, and this, in turn, by the direction of the stronger winds during the 40 days while the mackerel were passing through these miles phases of development and growth. These drifts may be as fast as 6 nautical the per day and may convey the mackerel several hundred miles. After reaching on post-larval stage (10 mm. and upward) the movements are less dependent drift, and probably are considerably aided by the tiny fishes' own swimming efforts. The BIOLOGY OF THE ATLANTIC MACKEREL 191

average rate of movement is sometimes about 3% nautical miles per day and may at times, on the part of the largest individuals, attain eleven nautical miles per day. In 1932, the combined drift and swimming movements brought the larvae to the shores of Long Island and southern New England. MORTALITY

Outstanding in the early life history of marine fishes is the high mortality in early stages. At sea, this is evident from the low numbers of larvae compared to the high number of eggs taken in plankton tows. In marine fish hatcheries, it has been evident from the high loss of larvae in all attempts to keep them beyond absorption of the yolk sac. It is probable that the fish cultural experience led to the generally accepted theory that the time of yolk sac absorption is the most critical period, and that it is so because the fish at that time must find proper food or die as soon as all the yolk is gone. Moreover, Hjort (see p. 207) believed that annual variation in the times and places of plankton increase during spawning might be such that an abundance of the right kind of food might coincide with this critical stage in one year and not in another. The coincidence of the two would produce a successful year class; the non-coincidence, a failure. However elaborate the theory, it has yet to be proved at sea that the yolk sac stage is critical or that the annual variation of mortality in this stage is responsible for the variation in year-class strength. Thus, a determination of mortality of the young stages of mackerel in 1932 is not only of interest in itself, but has an important bearing on the general theory of fluctuations in fish populations. Inasmuch as the year class of 1932 has subsequently faded to appear in the commercial stock in important numbers (Sette, 1938), the present examination of mackerel mortality in the season of 1932 deals with the record of a failing year class and should bring to light the stages that were critical in its failure. Determination of mortality.— There is at hand a simple way of determining the mortality rate of that year if it may be assumed that all the various egg and larval stages were sampled in proportion to their abundance in all parts of the spawning grounds, and during the entire period of planktonic existence. Then a frequency distribution of the summed numbers at each stage through the season would express their average relative numbers and constitute a survival curve. Although the sampling in 1932 approached a stage of perfection warranting treatment based on this general plan, there were nevertheless imperfections requiring secondary modifications, as will be explained. The actual drawing of hauls appears to have been qualitatively and quantitatively adequate. At each station, all levels at which eggs or larvae might be expected to occur were sampled uniformly, and the subsequent adjustment for volume of water strained per meter of depth provided totals at each station which may be taken as the summation of individuals below 17.07 square meters of sea surface, irrespective of their level in the water. Comparison of 1-meter and 2-meter net hauls indicated that there was relatively little selective escapement from the nets (p. 215). Also, the towing stations formed a pattern reasonably well covering all parts of the important spawning grounds off the United States coast. On the other hand, in some respects the samples did not adequately cover the entire season. At the time of the first cruise, spawning had already begun and larvae were taken for which there were no corresponding eggs. Simdarly, force of circumstances prevented cruises from being taken as frequently in July as earlier in 192 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

the season, and also prevented their continuation into August. Thus, there was less opportunity for taking large larvae corresponding to the eggs and small larvae of the earlier cruises. However, the cruises did thoroughly cover the major portion of the season of maximal spawning and subsequent larval development; so there need be only a treatment which excludes from comparison the large larvae early in the season and the eggs and young larvae late in the season which were not proportionately represented in the other stages of their planktonic existence. This was done by taking the average numbers of eggs and larvae per cruise for the several cruises that spanned the period of maximal numbers of each stage of egg 19 and larva. The selection of cruises for these averages was as follows: for egg stages A to C, cruises I to IV; 3-mm. larvae, cruises II to V; 4- to 7-mm. larvae, cruises III to VI; 8- to 9-mm. larvae, cruises IV to VII; 10- to 12-mm. larvae, cruises V to VIII; 13- to 15-mm. larvae, cruises VI to IX; 16- to 22-mm. larvae, cruises VII to IX; and 23- to 50-mm. larvae, cruise IX. BIOLOGY OF THE ATLANTIC MACKEREL 193

taken on the first 4 cruises, the treatment includes the population resulting from the course major portion, perhaps 70 percent, of the season's spawning. It of ignores the fate of the fewer eggs spawned prior to or later than the first four cruises, but the of the whole neglected portion is probably so small that it is unlikely that the survival season's brood of young differs from that of the treated portion. It could do so only included if the mortality of the neglected portion differed widely from the portion. There appears to be no reason for believing that there was any such wide difference. On the contrary, examination of the relative numbers of the various stages and sizes caught on those cruises which included a part of the history of the neglected portions suggests that these had a survival rate similar to that of the included portion. Having the average relative numbers of each category of egg and larva from this there remained the selected series (table 7, column headed "Average per cruise") in the duration of necessity of adjusting the numbers to compensate for the differences EGG LARVAL LENGTHS MILLIMETERS. ABC 5 6 7 8 9 10 15 20 25 30 40 50 T T nr T

- 100,000

10 ZO 30 40 50 ASE. DAYS.

Figure 17.—Survival of young stages of mackerel In 1932. Solid dots represent the means of three or four cruises each. Open a of survival circles represent the less reliable values based on only one cruise. The heavy lines represent simple interpretation fitted to the solid dots the method of rates, and the fine lines, a more complex alternative interpretation. Solid lines are by least squares. Lines of dashes connect their ends, and the line of dots and dashes is an extrapolation.

time represented by each egg stage and each larval-length class. The stages or classes catches of representing a long period of development would be passed slowly and the such a category would represent a larger accumulation of individuals than a category would be representing a shorter period of development. Since the accumulation relative values were directly proportional to the duration of the category, the true obtained by dividing the numbers of individuals in each category by the number of days required to pass through that category, according to the schedule, given in the column headed "Duration of category" of table 7. This, in effect, reduces the data to sub- to represent what the relative numbers would have been had it been possible— divide the material into categories that occupied uniform time intervals in this 194 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE instance, one day. The resulting values are given in the fourth column, and the logarithms of these (column 5) of table 7 are plotted in figure 17. Reliability ojihe survival curve.—The determination of the survival curve was based on plankton hauls generally considered to be only approximately quantitative, it utilized only selected portions of the original material, and it involved extensive computations. The reliabdity of the result therefore depends not only on quantitative adequacy of the original material, but also on whether the subsequent procedure introduced any biasing influences. The following discussion will draw attention to the facts which appear to have an obvious bearing on reliability. Unless some pertinent features have escaped notice, the conclusion is inevitable that this survival curve has surprisingly high reliability for all stages up to the length of 22 mm., or, for the first 60 days of life. Considering first the collection of material, attention may be confined to those influences that might possibly cause large larvae to be caught in relatively greater or lesser proportion than small larvae, for it is only by such "size selection" that the slope of the survival curve, and hence the conclusions as to mortality rates, could be affected. On this score there are two possibflities: the nets' catching ability might differ for different sizes of larvae; or the distribution of the larvae might vary in such a way as to cause a less complete sampling of one size than of another. In the appendix (p. 215) there is given evidence which appears to be indicative, if not conclusive proof, that the nets caught practically all the larvae in the paths of their travel, at least up to the 22 mm. size; hence net selection was probably not a biasing influence in this size range. Since the nets were fished from surface to below the thermocline, and since the larvae probably do not descend below that point (p. 173), and since straining was substantially uniform for all levels fished, there is little likelihood that differential vertical distribution was a biasing factor. There remains, then, the possibility that larvae of different sizes had different horizontal distributions, and that these distributions differed in a manner which would have affected the relative adequacy of the sampling of the various sizes. For small larvae up to 10 or 12 nun. in length, the drift was determined (pp. 183 to 191) with sufficient precision to establish the fact that the population of these sizes did not drift out of the area sampled. The majority of large larvae 22 to 53 mm. long, however, taken off eastern Massachusetts on the final (ninth) cruise, were outside the area covered on earlier cruises. Could, then, a portion of the population of medium sizes (12 to 22 mm.) have left the waters south and west of Nantucket thus have been under- Shoals, that is, the area of survey, prior to the ninth cruise, and sampled? If so, they should have been found in the intervening area during the eighth after cruise, which, fortunately, included that area. This cruise took place shortly the main portion of the larval population was in the 12- to 22-mm. size range. It included stations around Nantucket Shoals and on the portion of Georges Bank just M east of the Shoals; hence, in the area through which larvae would have been drifting or swimming if they had, by this time, begun their movement north and east past the Shoals. Since no larvae of these sizes were taken there, it seems unlikely that these sizes were undersampled as a consequence of emigration from the area south and west of the Shoals. In other words, the intermediate, as well as small sizes of larvae, were sampled in approximately then- true proportions.

M These stations of cruise VIII have not been included in any of the tables because the hauls there lacked pertinent material. BIOLOGY OF THE ATLANTIC MACKEREL 195

For the larvae over 22 mm. long there is no evidence to determine whether or not they were caught by the nets in their true proportions. On general grounds, one would expect that they could elude the nets, though the taking of a specimen as long as 51 mm. shows that the gear could catch at least some large-sized larvae. Offsetting the probability of undersampling the larger sizes, there is the opposite probabdity of over-

sampling them, because the stations were somewhat more closely spaced (see fig. 14) in the area north and east of Nantucket Shoals, where they were found, than south and west of the Shoals, where the smaller sizes were most abundant. Whether or not the loss of large larvae by eluding the nets and the gain by possible oversamplins: as the result of closer station spacing offset each other perfectly is indeterminable from the available data. Hence, the mortality determination is of uncertain reliability for sizes over 22 mm. For those smaller than 22 mm., the determination is reliable as far as collecting methods are concerned. Having found little reason to suspect size-connected biases in collecting, excepting possibly for sizes over 22 mm. long, two questions remain: were the hauls themselves sufficiently quantitative to give reliable indices of abundance for each station; and were the stations spaced properly to give a reliable summation of abundance for the entire area? To answer the first question separately would require a study of the variation in series of duplicate hauls, and is precluded for lack of material, but both questions may be answered simultaneously by a study of the relative numbers caught at the various stations in relation to the probable nature of distribution of numbers of individuals in the sea.

Inspection of charts of egg or larval distribution (fig. 13) suggests that the pattern of concentration has a form closely related to a normal frequency surface. Near the middle of the area in which eggs or larvae occur are one or several stations with very high concentrations corresponding to the mode; surrounding these are more stations with decidedly lower concentrations corresponding to the slopes; and at the periphery are many stations with very low concentrations corresponding to the "tads" of the normal frequency surface. Let us assume, for the moment, that the concentrations of eggs really do form a normal frequency surface. Then the number of a particular stage caught during a particular cruise is a reliable index of the abundance of that stage at that time, provided that: the stations where the catches were made were so located as to give proper relative representation of the various parts of this normal frequency surface, such as the mode, slopes, and tails; and that the catches also were sufficiently reliable to provide the true relative numbers to be found at the various parts of this surface. Therefore, a test as to the conformity of catches to the normal frequency surface would at once indicate whether the above assumption is correct; whether the catch stations were arranged so as to sample adequately the various parts of the distribution; and whether the hauls themselves were quantitatively reliable. To translate the normal frequency distribution into a convenient form for making 21 the tests, table 8 has been prepared. It was derived from the curve of the normal frequency distribution where, for unit standard deviation and unit N

y=0.3989e 2

11 Buchanon-Wollaston (1935, p. 85) has given a table purporting to give the same statistics, but it appears to represent the results of sampling only along a line passing through the mode, of a normal frequency surface, not the results of sampling over the entire surface. For the latter, account must be taken of the fact that in such a surface, so sampled, the areas of classes of equal range in ordinate height increase as the square of the distance from the mode. 525293—44 4 196 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

2 by calculating for values of y (catch magnitudes) the corresponding values of x (relative number of catches) over a range of y from 10,000 to 5 and at intervals of 500 for the first 19 classes, of 25 for the next 19 classes, and of 5 for the next 4 classes. 2 For convenience the x series was converted to values giving a cumulative total of approximately 1,000 (actually 999.96). This table can be used for any range of catch sizes in which the maximum is not more than 2,000 times as large as the minimum, by first multiplying the empirical values by 10,000 times the reciprocal of the maximum catch. Linear interpolation is fairly accurate in the table ranges of 10,000 to 5,000 and of 500 to 250; but the work is facilitated and is more accurate for all parts of the range when the tabular values are graphed.

Table 8. —Relative number of catches of given magnitudes to be expected from a •population of organisms distributed in the form of a normal frequency surface

Magnitude of catch BIOLOGY OF THE ATLANTIC MACKEREL 197

Table 9. —Example of the compulation of limits for 5 classes within each of which an equal number of catches would be expected if the distribution of stage A eggs during cruise I conformed to a normal frequency surface; and the actual and theoretical number of catches for these class limits

1 198 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 10. —Summary of test to determine whether the magnitudes of catches of eggs and larvae conformed to the distribution expected from sampling a normal frequency surface

Stage BIOLOGY OF THE ATLANTIC MACKEREL 199 by chance, produce a series of catch magnitudes conforming so well to hypothesis, it has been proved not only that the hauls were quantitative, but also that the samp- distribution of each of the ling provided adequate representation of all parts of the various stages of eggs and larvae up to 22 mm. long. Nothing is yet proved as to the extent of random variability, either of the quantities caught per haul or per cruise. This would control the scatter of points in figure 17 and will receive consideration in the final paragraphs of this section. The foregoing has dealt with the collection of material. Turning now to the mathematical treatment, the initial step was to total the catches of a given stage for each cruise and then average these totals for certain groups of cruises. This use of total per cruise is equivalent to a direct arithmetic integration of the frequency surface and could introduce no errors if the same stations were occupied on each cruise, and if all stations represented equal unit areas. These requirements were approximately the stations met because the same station plan (fig. 14) was used for each cruise, and were distributed uniformly enough to represent approximately equal unit areas. The principal change from cruise to cruise was the omission of some stations. As earlier mentioned, stations north and east of Nantucket Shoals were omitted from the no first seven cruises, and it already has been pointed out that this probably had effect on the computation because these northeasterly stations could have contributed nothing to the totals of the group of mackerel that is followed in the survival curve. Besides this the stations at Martha's Vineyard IV, Montauk IV, New York V and the VI, and Cape May I were usually omitted. Since they proved always to be at periphery of the egg and larval concentrations, their exclusion or inclusion could make little difference. However, on four of the first seven cruises, there were additional omissions which could possibly have had important effects. On cruise I the station at New York I and all of those on the Montauk and Shinnecock sections were omitted. Judging from the catches at adjacent stations, and also from the distribution of appropriate stages on the following cruise, tliree of these omitted stations might have added low to medium catches to the totals for that cruise stage A and B eggs, but this could not have increased their totals for by four more than 5 percent, and could have modified the average per cruise of the cruises used for these stages by less than 2 percent, so the effect of this omission is inappreciable. On cruise IV all stations on the Winterquarter section, and those at Chesapeake I and III were omitted. This omission would have a serious effect on the total for that of cruise, for these stations could have been expected to yield nearly maximal numbers 4- to 8-mm. larvae, but the effect of this omission was rectified by substituting the cruise III values for these stations in calculating the average per cruise. (See foot- extent of 2 note p. 192.) This substitution could have introduced error only to the days' growth and mortality—an effect that would not be perceptible after inclusion of the data for the three other cruises in the group average. On cruise V the stations on the Martha's Vineyard section, at Montauk III, and at Shinnecock I and II were omitted. This probably reduced the totals of 3-mm. larvae is larvae appreciably, and 4-mm. larvae slightly. If the effect on the 3-mm. from the estimated by examining the result of substituting numbers interpolated larvae is previous and subsequent sampling at these stations, the total for this size of increased from 5,215 to 12,549 for cruise V and the average per cruise for cruises II to V is increased from 9,310 to 11,144. Substituting the latter in table 7 and carrying the work through to the logarithm of the empirical number surviving per million, it is 200 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE found that the value increases from 5.299 to 5.378, indicating that the point for 3-mm. be raised an amount to the diameter of in figure 17 should probably by nearly equal catches of 4-mm. larvae indicates the dot representing it. Similar examination of that the total for cruise V might be raised from 8,236 to 9,945, a change that becomes to the values on the of survival. imperceptible when worked through graph were On cruise VIII the stations at Fenwick, Winterquarter, and Chesapeake of the omitted. At the very most these could have contributed nothing to any a few individuals in the averages involving this cruise, excepting possibly very 7-, would not cause a in the survival 8-, and 9-mm. classes. These perceptible change curve. the New York By the time of cruise IX, only one larva was found along section, south of that that the omisson and it was so probable that none at all remained locality effect on the survival of all stations from there southward could not have had any curve. introduced no errors other Hence it may be concluded that the use of cruise totals on the survival curve. than a slight lowering of the 3-mm. point introduced the selection of Turning now to the possibility that errors were by on the survival curve con- certain cruises, it will be recalled that the successive points later cruises for sist of averages of the catches in groups of cruises, using successively the season the successively older larvae so as to follow the main population through to of boat from egg stages to late post-larvae. Owing, however, exigencies operation, so the cruises toward the end of the season were separated by wider intervals of time, from more that the average numbers of older larvae were calculated samples widely more submaximal values for spaced in time. This would tend to include relatively the older larvae than were included for the eggs and younger larvae. Although the extreme amount effect of this cannot be directly measured, it is possible to deduce the of distortion to be expected from the inclusion of submaximal values. of the This can be done by restoring submaximal values to the computation for 5-mm. the of the average number of young larvae. For instance, larvae, average catches for cruises III to VII, which were the ones used in the mortality determina- and VII would restore submaximal values tion, was 1,760. Inclusion of cruises I, II, in column 3 of table and produce an average of 1,220. Substituting the latter figure a of 4.387 instead of 7 and carrying the computation over to column 5 gives figure 17 4.547 for the 5-mm. class. This would lower the point for 5 mm. in figure by This about 1H times the diameter of the dot representing that point in the graph. increase in submaximal is a very small alteration brought about by a relatively large of values. Therefore the inclusion of what was probably a relatively small number submaximal values for the older larvae by the method used in averaging cruises to the older obtain the mortality curve could have lowered the points representing curve the larvae very little indeed, and therefore have altered the by only slightest amount. Next may be examined the distortion that could be connected with the growth- rate data employed in computing the mortality curve. Evidences of the reliability section on that and it of the growth-rate determination were given in the subject, was concluded that the general course of the growth curves must be essentially correct. It remains to be considered here whether there might nonetheless actually not reflected in the have been irregularities in growth, and because they were growth the observed statistics used in computing mortality rates, they could have produced peculiarities in the survival curve. BIOLOGY OF THE ATLANTIC MACKEREL 201

The outstanding peculiarity in the survival curve is, of course, the abrupt change of level and slope at the age of 40 days, or length of 10 mm. To investigate effect of a fluctua- the possibility that this might have been due to the mathematical tion in growth rate, rather than a fluctuation in mortality rate, let it be assumed that the mortality rate through and beyond this period was constant, and compute the changes in growth rate required to fit this hypothesis. The resulting new values for growth rate, in terms of days required to grow one mm. in length, are as follows: Millimeters: -Dai" Millimeters—Continued. Day 9 3.04 13 15 10 --- .80 14 .18 11 .38 15 09

12 . 24

Thus, this hypothesis would require growth at an ever-accelerating rate from 10 mm. on, such that less than a day would be occupied in growing from a length of 10 to a length of 15 mm., and by that time growth would be at the rate of 10 mm. per day. Clearly this hypothesis is untenable, for such high growth rates are not only of larvae absurd per se, but also inconsistent with the distributions of lengths taken on successive cruises; and it may be concluded that the outstanding peculiarity in the mortality curve cannot have resulted from a fluctuation in growth rate. This demonstration, having proved that it requires striking changes in growth rate to produce material effects on the survival curve, indicates also that errors of the order of magnitude which likely exist in the determination of growth would not materially affect the determination of mortality rates. Thus far attention has been centered on the possible elements of selective error or bias connected either with collection of the material or the subsequent mathematical treatment. There remains the question of the effect of random variability. This could not alter the level or the trend of the survival curve, for random variability would produce empirical values that tend to deviate equally above and below the true values, so that the sole effect would be on the scatter of points, or, in other words, the relative reliability of fit by any lines expressing their trends. This is readily investigated by conventional statistical methods. Because the points in the curve obviously lie along straight lines over consider- to various able segments, such lines have been fitted, by the method of least squares, combinations of segments. Since our interest hes principally in the mortality rates or expressed by the slopes of the lines, attention may be focussed on the 6 value, 22 regression coefficient, in the equation: y=a-\-bx which describes these lines. The standard deviation s of the regression coefficient b may be estimated by the formula _S(y-D 2 1 tin —2 of the To investigate the reliability of the slopes of the lines for various segments diagram, one may calculate

, Jb-Pi/S(x-x)* 8 and find, from published tables, the probability, P, that any other slope /3 might result from sampling the same universe. Being interested in knowing the limits of

» The symbols given In this and following equations are those used by Fisher (1932). 202 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE accuracy of the slopes, values of t may be selected for P=0.05, and by substituting these in the equation, st

values of 6-/3 may be calculated which, when added to b, or subtracted from it, will give the limits of a range of slope values. The chances will then be 19 out of 20 that the true slope lies within this range. From these calculated ranges (table 11), it is clear that there was so little random variability of the points about the lines of best fit, that mortality values are accurate to within one or 2 percent per day for all segments other than A to C. There still remains the question: which of these combinations of straight lines gives the most probably true series of survival rates? This may be investigated by the formula for the significance of the difference of two slopes, again going through the t test, using the formula — 6i 62 t =

V^Lsfe-xo^sfe-^) 2] where

^_~ S(y,-Yl y+S(y2 -Y2y S n'-4

the results in table 1 1 where the of b initial From given , subscripts represent the and terminal points of the segments, it is apparent: (1) That &a-c differs from 6 4 _ 8 just enough to indicate that the survival rate probably is significantly higher in the larval stages than in the egg, and therefore the two lines A-C and 4-8 better describe this segment than the one line A-9. However, the latter does not differ significantly enough from each of the former to preclude the possibility that it fairly well represents the general course of survival from the early egg stage to the 9-mm. larva.

is . (2) That 6n_22 certainly significantly different from 6a-9, though not from 6 4 _ 8 The interpretation of these findings will be discussed in the following section.

Table 11.—Estimates of accuracy of slopes of lines in figure 17

Segment BIOLOGY OF THE ATLANTIC MACKEREL 203

Mortality rates.—When the logarithms of the fully adjusted survival numbers are plotted, as in figure 17, the series describes Dearly straight lines over certain portions of its extent, indicating that in each of these straight-line segments, mortality must have proceeded at a uniform percental rate. The major feature to be noted is the break at about 35 days when the larvae are 10 mm. long. At this point there is a change of level and of slope which may be considered as dividing the curve into three portions: (1) egg, yolk-sac, and larval stages, (2) transition between larval and post- larval stages, and (3) post-larval stage. Each will be discussed separately. The first portion representing stages up to 10 mm. in length is subject to alternative interpretations due to the nearly, but not wholly, linear arrangement of points. The simplest interpretation is that the mortality rate was uniform and that the deviations from linearity were due to defective sampling. If so, the single heavy straight line drawn from A to 9 mm. in figure 17 expresses the mortality. Accordingly, this mortality was at a constant rate, and amounted to 14 percent per day. On the other hand, it has been shown in the previous section that there is little ground for sus- pecting serious defects in sampling, and also that the slope of the line A to C differs significantly from that of the line 4 to 8 mm. This being true, the mortality rate would be better described by the three fine lines of figure 17, the one extending from A- to C-stage eggs; another from 4- to 8-mm. larvae; and still another joining their ends across the 3-mm. (yolk-sac) stage. According to this interpretation, the initial

rate, i. e., the rate during the egg stage, was 5 percent per day. The next rate, i. e., during the yolk-sac stage, was 23 percent per day, and the third rate, i. e., during the 23 larval stage, was 12 percent per day. However, according to both interpretations, mortality has reduced the population to about one-tenth of its original numbers by the time the larvae reach 4 mm. long, and when they attain 9 mm. in length at 35 days of age, to one-thirtieth of the original number. If any one period is to be singled out as the most critical, it must be the ensuing period during the transition from larval to post-larval stages, when in passing from 9 to 11 mm., the numbers are reduced by 90 percent in the short space of about 3 days. The rate of mortality may be variously computed, depending on the choice of straight lines in figure 17. The lowest is 30 percent, and the highest, 44 percent per day. Either of these rates is distinctly higher than the highest alternative estimate (23 percent per day) in the yolk-sac stage. The high mortality during this short period, coupled with the losses previous to this stage, reduced the survivors to only one three-hundredth of their original numbers; thus the population was already severely decimated on entering the post-larval stage. During the post-larval stage, the rate of mortality apparently was more moderate than in earlier stages. The data on which the rates are based appear fairly reliable up to the 22-mm. stage, or 62 days of age, and the fitted line for the segment 11 to 22 mm. in figure 17 represents a mortality of slightly over 10 percent per day. Beyond 22 mm. the catches of larvae were few and were confined to only one cruise, so that the reliability of their relative numbers is in doubt; but the evidence, such as it is, points towards the continuation of the same rate of mortality to the size of 50 mm., or age of 85 days. Restating the history of mortality, it appears that there was a general basic rate of 10 to 14 percent mortality per day throughout the period studied. The most important deviation from this general rate was during the 9- to 11-mm. stage, when the population suffered about 30 to 45 percent mortality per day. Other deviations " Also, according to this interpretation, the data In the last column of table 7 should be taken as representing the number of survivors per 840,000 newly spawned eggs Instead of per million, as given In the column beading. 204 FISHEKT BULLETIN OF THE FISH AND WILDLIFE SERVICE of somewhat doubtful significance occurred during the egg stages, when a lower rate of 5 percent per day was indicated, and during the yolk-sac stage, when a higher rate of about 23 percent may have intervened. The net survival to the 22-mm. stage, or 62-day age, was 40 per million newly spawned eggs, and, assuming a continuation of the 11 to 22-mm. rate of mortality to the 50-mm. stage, or 85-day age, it was 4 per million newly spawned eggs. Discussion.—Since it is probable that the success or failure of year classes is determined during early life, and since it is known that the year class of 1932 was a failure, it is natural to assume that the mortality curve just given represents the record of that failure. That this is true appears from the following considerations. From fecundity data (p. 156) it is estimated that a female spawns about 500,000 eggs per year, and from the size composition of the adult stock (unpublished notes) it may be estimated that each female spawns over an average period of about four years, producing a total of 2,000,000 eggs. Therefore, to keep the population constant, from

2,000,000 eggs, one female on the average should reach average spawning age; i.e., a survival of one fish per million. But in 1932 only four fish per million were left at the early age of three months. At this age, the rate of mortality was about 10 percent per day. Were this rate to continue only 35 days longer, the survivors would number only 0.1 per million; i.e., only 0.1 the number required to reach average spawning age. Of course, it should not be assumed that the 10 percent mortality would continue indefinitely. But even should it be as low as 2 percent per day, the year class would be reduced to the 0.1 per million level before the end of the first year of life; and even then they are at least 2 years removed from average spawning age. To reach that age with survival of one per million, mortality could not average more than 0.12 percent per day during the time intervening between 50 mm. and average spawning age. It is unreasonable to suppose that the mortality, last observed at 10 percent per day, coidd immediately drop to such a low rate and remain there. Hence it is likely that a year class, to be successful, must have a survival well above four per million at the 50-mm. size, and that the 1932 class was a failure because of the high mortality during stages preceding the 50-mm. length. The causes of this failure may be sought in the record of mortality during the various stages. The outstanding feature in this record is that no single period could be considered crucial in the survival of the year class with which we are concerned. Mortality in all phases of development contributed substantially to the decimation of the population. This fact is most readily appreciated when the contribution to total mortality by the periods of relatively high rate is compared with the contributions by the periods of low rate. The mortalities in the yolk-sac stage and in the transition between larval and post-larval stages (taking the highest alternatives in each case) together represent the passage through 1.9 logarithmic phases. All the other stages together represent 3.6 logarithmic phases. Hence, one may say that about one-third of the mortality was suffered during the so-called "critical" stages, and the other two-thirds during what might be called "non-critical" stages. The question naturally arises, which of these was in 1932 the determining factor in the failure of the year class? To answer the question calls for comparable data on mortality during the early life history of a successful year class. Lacking this, one can only speculate. If in 1932 the so-called critical stages were to have been eliminated, the survival to the 50-mm. point would have been 250 per million eggs spawned. If the so-called noncritical stages were to have been eliminated, it would have been 12,500 per million eggs spawned. Of course, it is difficult to conceive of complete BIOLOGY OF THE ATLANTIC MACKEREL 205 elimination of mortality from any of these stages, but if a year class is to be successful there is obviously greatest opportunity for improved survival in the noncritical stages, for they contributed most heavily to the failure of the year class. For tbis reason, one must look with at least as much suspicion on the mortality during non-critical stages as on the mortality during critical stages when in search for casual agencies that may have been operative during 1932. In looking for such agencies, there are two features of the 1932 season that appeared to be unusual and of the sort likely to have affected survival. One of these was the relative paucity of zooplankton in the area of survey during the spring and early summer (i.e., May and June). The zooplankton catches averaged only 280 cc. per haul, as compared with 556 cc. in 1931 and 547 cc. in 1930 (Bigelow and Sears, 1939, p. 200). Both of the last named seasons produced good year classes, and there is, therefore, an indication of correlation between zooplankton abundance and the survival of a mackerel year class. If failure to survive in good numbers in 1932 was in fact due to dearth of food, and the dearth was continuous throughout the season 206 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE and Hodgson (1937) reported correspondence of relative success of six herring year classes and the strength of winds from certain quarters as inferred from pressure gradients; and Carruthers (1938) amplified these findings, presenting the relation for 11-year classes in the East Anglian herring fishery. He concluded: "It is reasonable to argue along these lines:—as from year to year, increased 'from-Channel' air flow means increased 'from-Channel' water flow, and this in turn means:— (1) Tbat the passively drifting spawning products will be drifted farther afield—apparently * * a good survival augury for the herring *." In the same paper, Carruthers demonstrated the parallelism between changes in both certain pressure gradients and east wind component, on the one hand, and relative strength in a series of 15 haddock year classes on the other hand. These illustrations support the theory that local winds affect year-class survival. Though they demonstrate the importance of transport, the remainder of the survival (or mortality) mechanism, particularly its biological aspects, has yet to be elucidated. For the mackerel of the American Atlantic seaboard, however, it is possible to advance a reasonable explanation for the connection between wind direction and survival. The center of spawning, it will be remembered, is southwest of Fire Island. The juvenile nursery grounds, judging from relative quantities of young mackerel usually found along various parts of the Atlantic seaboard, is along the coast of southern New England from Cape Ann to about the eastern end of Long Island. Therefore the prevalent southwest winds during May of 1930, 1931, and 1933 con- veyed the larvae toward the nursery grounds. Conversely, the prevalently northeasterly winds of May 1932, on the average, were of hindrance rather than help to the larvae in reaching their nursery ground. If this be true, there is the further probability that the significantly higher mortality in 1932 at the transition phase when fins were developing was a consequence of the pattern of drift in that year. The formation of fins and their subsequent use undoubtedly enlarged the expenditure of energy and hence increased the food re- quirement at the transition phase. At this time, on the average, the larvae were still distant from their nursery ground and if feeding was poorer where they were than on the nursery ground, the observed heightened mortality at this phase would thus be explained. Shortly after, by directional swimming, and with some assistance from favorable winds, some of the larvae did reach the presumedly more favorable location and thereafter were subject to a distinctly lower mortality rate. Thus, there are evident two influences that contributed to the failure of the 1932 class. One was the general paucity of plankton, which probably increased mortality throughout the entire early life history; the other was the apparently unusual direction of their drift, which probably heightened mortality mainly during the transition from larval to post-larval stages. Though either one of these influences might con- ceivably have been the sole cause of the failure of the 1932 class, the shape of the survival curve suggests that both contributed substantially. Indeed, the two might be related to each other as well as to the mortality of the mackerel. To be sure, these are speculative conclusions. However, they furnish hypotheses that should be useful in planning further observations, especially in seasons of successful survival. Significance of observed mortality in 1932.—Although one season's observations on one species of fish form a slender basis for generalizations, the fact that it is perhaps the only determination of mortality of a marine species under natural conditions gives special significance to the results, for it affords opportunity, for the first time, of comparing actual observations with theory. BIOLOGY OF THE ATLANTIC MACKEREL 207

In stating existing theory, one can do no better than to quote Johan Hjort, who, perhaps more than anyone else, was responsible for bringing attention to the importance of year-class success or failure as the explanation of fluctuations in the sea fisheries. In 1914 he advanced, and in 1926 (p. 32) reiterated, the theory that:

The rich year-classes appear to make their influence felt when still quite young; in other words, the numerical value of a year-class is apparently determined at a very early stage, and continues in approximately the same relation to that of other year-classes throughout the life of the individuals. It has already been shown that the observations on mackerel in 1932 are in harmony with this theory (p. 204). Hjort (1926, p. 33) in discussing the great Norwegian cod and herring fisheries, suggested further: As factors, or rather events which might be expected to determine the numerical value of a new year-class, I drew attention to the following two possibilities: (1) That those individuals which at the very moment of their being hatched did not succeed in finding the very special food they wanted would die from hunger. That in other words the origin of a rich year-class would require the contemporary hatching of the eggs and the development of the special sort of plants or nauplii which the newly hatched larva needed for its nourishment. (2) That the young larvae might be carried far away out over the great depths of the Norwegian Sea, where they would not be able to return and reach the bottom on the continental shelf before the plankton in the waters died out during the autumn months of their first year of life.

Observations on mackerel do not support the first possibility. Mortality immediately after hatching was little, if any, greater than at other times, and hence failure of the 1932 class could not have been due to acute dearth of food at the hatching time. If shortage of food was responsible, it had its effect either throughout the period of planktonic existence or at the transition phase (9- to 10-mm.), well after the hatching time.

On the other hand, the second possibility has strong indications of support in the mackerel data. Not only did the heightened mortality at the 9- to 10-mm. lengths appear to be connected with drift of the larvae, but there also was a marked correspondence between success of the year-classes 1930 to 1933, and the drift that they must have experienced as the result of dominant winds in May of these four years. That drift may in general be an important influence on success of year classes is further suggested by a similar finding for the American haddock (Walford, 1938, p. 55), wherein the relative failure of the 1932 class corresponded with drift of larvae away from Georges Bank, and relative success of the 1931 class corresponded with a pattern of circulation that kept the larval population on Georges Bank. Thus, in the two instances where the events at sea have been traced, it was the oceanic circulation that influenced the success of year-classes; and in the one case where the course of mortality (in a failing year class) at sea was traced, it was not any, if at all, higher at the hatching time, and hence failure could not be attributed to acute shortage of food at this period. In addition to the actual facts observed and their contribution to the understanding of year-class success or failure, the development of technique for determining mortality rates can have significant influence on future development of fishery science. If applied over a series of years, it woidd provide the data needed for separately evalu- ating the correlation of the size of the spawning stock with numbers of resulting offspring, and the correlation of the survival of offspring with the contribution of the year-class to the commercial stock. The predictive uses of such knowledge would be of obvious value to the conduct of fishing operations and to the trade in fishery products. But the value of such knowledge in formulating conservation policies would 208 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE be even greater than its value for predictions. These separate correlations would provide a basis for determining the size of spawning stock necessary to maintain an undepleted fishery. Efficient utilization will be possible when a reliable estimate can be made of tbe proper size of spawning reserve. Until then, there will always be danger of reducing the annual take, on the one hand, by attempting to preserve more spawners than needed, or, on the other hand, by catching more spawners than can be spared from the stock needed for adequate reproduction. APPENDIX METHODS OF DETERMINING SIZE AT MATURITY

Samples of fish were taken at various times at Woods Hole, Provincetown, and Sagamore, Mass., during the period June 24 to July 21, 1925. The fish were measured to the nearest half centimeter on a straight line from tip of snout to the extremity of the midcaudal rays. Gonads of the males were graded by eye as small translucent, small opaque gray, enlarged white, running milt, and spent. The last three grades were classified as mature. Gonads of females were graded by eye as small translucent, small granular, enlarged granular, translucent spots, running ripe, and spent. The last three grades were classified as mature. The results are summarized in table 13.

Table 13. —Size of mackerel at maturity as indicated by 1,116 individuals taken by traps in the vicinity of Woods Hole, Mass., and in Massachusetts Bay during the period, June £4 to July SI, 1925

Length, centimeters BIOLOGY OF THE ATLANTIC MACKEREL 209

Table 13. —Size of mackerel at maturity as indicated by 1,116 individuals taken by traps in the vicinity of Woods Hole, Mass., and in Massachusetts Bay during the period, June 24 to July 21, 1925.— Continued.

Length, centimeters 210 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

45° above the first net and 28° below it. The depth ranges of fishing would then be 0-18 meters and 22-44 meters, respectively, for the upper and lower nets. Since the course of plankton nets through the water usually is undulating (Russell, 1925, pp. 603-604), the theoretically unfished gap between the nets and the theoretically stepwise character of hauls would both be practically obliterated and the sampling virtually uniform, except for the greater depth range covered in unit time by the lower net. The latter was taken into account in the subsequent treatment of data. During the eighth and ninth cruises when the hauls were made with a 2-meter net, only one such net was used, and at the deeper stations it was sent down to a depth roughly equivalent to that reached by the deeper of the two nets employed on earlier cruises, so that the single, oblique haul of the 2-meter net sampled through approximately the same strata as the two nets of the preceding cruises. Measurement oj quantity of water strained by the nets. —It is obvious that two variables, speed of towing and degree of clogging, seriously modify the flow of water through plankton nets, causing variations in the catching capacity. To eliminate these sources of variability, a current meter was installed in the mouth of the net to measure the flow. The utility of current meters in measuring the volume of water passing through a plankton net depends on whether or not the flow past the meter is equal to or proportional to the average flow of water into the net. By towing, at usual speeds, a standard net with a current meter in the center of the mouth and another meter at the periphery, it was found (William C. Herrington, unpublished notes) that the flow past these two positions differed less than 10 percent. Since these positions were such as to register the maximum difference in rate of flow, if any ex- isted, this evidence was taken as indicating uniform flow into all parts of the mouth of the net. Hence we regarded the registration of flow past the meter as directly measuring the flow tlirough the entire opening. The instrument used for measuring the flow consisted of the propeller mechanism and revolution counter from a dismembered Ekmann current meter, turning five to six revolutions per meter of flow at usual towing speeds. For precise determination the meter was calibrated over the range of towing speeds. The total revolutions turned during a tow were converted to speed by dividing by the duration of the haul, in seconds; and the equivalent rates of flow were found from the calibration graph. These are the rates used in the specimen computation of table 15. While the current meter was used as a standard procedure, there were times when mechanical difficulties prevented proper registration. To provide basic data for comparable treatment of hauls made on such occasions, records were taken periodi- cally, during each haul, of the towing wire's angle of stray and of the ship's speed as measured by timing the progress of the ship past a chip cast alongside. An estimate of the extent to which the net was clogged was made at the end of each haul. Rela- tions between these observations and flow past the current meter gave average factors by which angles of stray or ship's speed could be translated to terms of equivalent current meter measurements. This afforded means of estimating the flow on those hauls which were not accompanied by reliable current-meter records. All the hauls class. of cruise I, and 5 percent of the hauls on subsequent cruises were of this For these hauls there was some error of estimate which may have been considerable for individual instances, but were, we believe, of random nature tending to balance each other, and so could have introduced very little inaccuracy into the general results, based on averages of a number of stations. BIOLOGY OF THE ATLANTIC MACKEREL 211

Only one current meter was available, and this was used in the upper of the two nets. When more than one net was on the line, the flow through the lower net was assumed to be the same as that through the upper net except as modified by clogging. Four degrees of clogging were recognized according to the following definitions: —When net is hauled to deck, water runs freely out of net and cod-end so that no water is left by the time the net reaches deck. 1 —-Water runs out of net freely but out of cod-end slowly so that some water is left in cod-end when net reaches deck. 2—Water runs out of net so slowly that it remains above level of cod-end coupling when net reaches deck, but falls to level of coupling after a short interval of time. 3—Entire net visibly covered with clogging organisms and water stays above coupling so that special means must be taken for washing down net. By the graphical partial correlation method (Ezekiel, 1930, pp. 143-145), it was found how much the relation between the angle of stray and the quantity of water strained was modified by the various degrees of clogging. The amounts by which the rate of flow for of was clogging— changed average given angles stray +0.032, —0.03, 0.073, and —0.108 meters per second for cloggings of 0, 1, 2, and 3, respectively, on the clogging scale as above defined. For the hauls made without current meters in the nets, these values were added to the theoretical flow as estimated from the angle of stray. The magnitude of these corrections is given by their percentage relations to the average rate of flow, which were +8, —1, —18, and —26 for the respective degrees of clogging. These, of course, are averages for each of the 4 degrees of clogging. The extreme individual values were plus 37 percent and minus 29 percent, which indicates that the total flow through an extremely clogged net at times was only half as much as through a very clean net. Since the clogging is progressive during a haul, it is obvious that practically no water is strained toward the end of any haul in which the net becomes badly clogged. The hauling method employed in tbis work, therefore, would undersample the upper layers relative to the lower layers. This would be a serious difficulty if clogging were often severe, but during 1932 only 4 percent of the hauls were of third degree and 15 percent of second degree clogging; hence uneven vertical distribution of sampling did not often occur. No adjustment was made for

trim fifi Apt ENUMERATION OF EGGS AND LARVAE

Eggs and small larvae were so abundant in many of the meter-net catches that a sampling method was necessary to estimate the total numbers caught. The formahn preserved plankton catch was transferred to a wide-mouthed graduated receptacle, enough liquid added to bring the level to a certain mark (often 2,000 cc), the contents stirred vigorously to mix uniformly, and a dipper then plunged into the mixture and withdrawn level full. The dippers were of the type made for dipping cream, each comprised of a small straight-sided cup with a long handle. Several sizes of dipper, each of known capacity, were used and one or several dipperfuls taken, depending on the size of sample desired. All fish eggs and larvae were removed from the sample. From the remainder of the catch, all larvae larger than about 5 mm. in length were removed. From the 2-meter net catches all the larvae were removed. Mackerel eggs and larvae were separated from those of other species and further examined, counting the number of eggs at each of three stages of development and the number of larvae at each millimeter of length. Measurements were made with the aid of microscope and eye-piece micrometer for larvae under 7 mm. and with millimeter rule and unaided eye for larger ones. The measurement was from tip of snout to end of notochord in larvae, and to base of caudal rays in post-larvae. Dis- 525293—44 5 212 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE torted specimens were classed by matching them with straight specimens of known length. The method of converting the counts to total catch was simple in the majority of instances because usually the mackerel material consisted either entirely of eggs and small larvae, so that the total catch could be computed directly from the known volume of sample sorted and the known volume of the plankton from which the sample was drawn; or entirely of large larvae sorted from the entire catch, so that a simple count represented the total. In a minority of instances, when both small and large larvae occurred in the same haul the total had to be computed from a combination of the sampled numbers of small larvae and the total numbers of large larvae. The specimen tabulation (table 14) illustrating the computation is self- explanatory except for the treatment of those sizes of larvae which were too scarce to be adequately represented in the small sample. Referring to columns 2 and 3 of table 14, it is obvious that the numbers of 8-mm. larvae were too few to have been taken in the smaU sample and also that in sorting the remainder, larvae as small as 6 mm. and perhaps also 7 mm. were not fully removed. Therefore, the 3- to 6-mm. larvae, inclusive, in the small sample were taken as representing the catch of these sizes and the items of column 2 were multiplied by ^r~r and entered in column 4. The numbers (2) in the 7-mm. category in the small sample (column 2) were taken as representing the numbers of larvae 7 mm. and over, which should then

2,000' total 2 X =36 in the entire sample. Since there were known to be 6 larvae of 8-mm. length (column 3) in the catch, the entry of 6 was made opposite the 8-mm. class in column 4 and the entry of 36 — 6 = 30 opposite the 7-mm. class. The count of larvae in the lower haul (table 14) included no larvae larger than those found in the small sample, and the total numbers of each size (column 8) were computed simply

' by multiplying the counts in the sample (column 6) by .

Table 14.—Specimen computation for converting counts of eggs and larvae to total catch on the standard basis of 17.07 cubic meters of water strained per meter of depth fished

[Data relate to station 21491] BIOLOGY OF THE ATLANTIC MACKEREL 213

In this particular sample the special treatment concerned the larvae of 7-mm. and upward. This was not uniformly true. The completeness of removal of large larvae from the remainder varied with the character of the plankton with which they were mixed and also, no doubt, with the fatigue of the person sorting the material. Due to this -variation each haul was treated according to the internal evidence provided hy the counts therefrom. More often than not the relative counts of the small sample and of the remainder indicated completeness of removal of smaller than 7-mm. sizes from the latter so that the length of larvae concerned in the special treatment was usually 5 or 6 mm. rather than 7 mm. as in the sample given. COMPUTATIONS OF CATCH PER STATION

Standard haul.—Since it was desired to have a number representing the total population of eggs and larvae at each station, regardless of depth, the catches were converted to the basis of a standard amount of straining per meter of depth fished. The standard amount selected was the average of actual performance, as measured by the current meter during the first seven cruises of 1932, which was 17.07 cubic meters of water strained per meter of depth fished. The average performance was taken rather than any arbitraiy amount because it involved a minimum alteration of original data, and the resulting figures represent nearly the actual numbers caugbt, except for the last two cruises, when the adjusted two-meter net catches represent approximately one-sixth of the actual numbers taken. Where an upper and a lower net were employed, the standardized catch of the lower net was added to the standardized catch of the upper net after a correction for contamination was applied to the numbers found in the catch of the lower net. The computations are illustrated in table 15. The procedure for 2-meter-net hauls was exactly the same as for 1-net hauls by 1-meter nets except that an additional factor of one-fourth was applied to offset the quadrupled cross-sectional area of the net's mouth. Other things being equal, this would have resulted in standardization factors about one-fourth as large as those for the 1-meter nets, but actually the 2-meter net was towed somewhat faster and its oblique path was somewhat more gradual due to a higher towing angle in relation to the amount of line hauled in at each time interval. Hence the average amount of water strained per meter of depth fished was about 6 times, instead of 4 times, as great as in the 1-meter nets, and the factors for standardizing accordingly averaged about one-sixth. For both sizes of net, therefore, the resulting factors for standardizing given in the columns headed "S factor" in tables 17 and 18 are such as to convert the catches at each station to the equivalent of the numbers that would be found in a column of water with a cross-sectional area 17.07 square meters, and extending from the surface to the deepest level reached by the nets at each station. This may also be stated as being equivalent to 21.7 times a vertical haul of a 1-meter net of perfect straining capacity. 214 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 15.—Specimen computation of factors for adjustment of haul to standard basis of straining 17.07 cubic meters of water per meter of depth fished and for ascertaining contamination of catch of the lower net in passing through the upper stratum

[The data relate to station 21491]

Item Unit Upper net Lower net

Length of line payed out Meters 0-25 25-55 Average stray Degrees from vertical. 51.3 35.0 Stratum fished Meters 0-16 20-45 Thickness of stratum fished Meters 16 16 Time fished (exclusive of time spent by the lower net in passing through Seconds 865 the upper stratum). Rate of flow through net (from current meter) Meters per second. 0.574 Clogging (on arbitrary scale, see text) 1 1 Adjustment for clogging Meters per second. -0. 007 Adjusted flow (item 6 plus item 8) Meters per second. 0.574 0.567 Total flow (item 5 times item 9) __. Meters 496 556

Standard flow 4 times 17.07— _ Meters 348 (item )

Factor for adjusting to standard haul . ) . 0.70 0.63 ( fl Time spent by lower net in passing through the upper stratum Seconds. 127 Flow through net while passing through the upper stratum (item 9 times Meters.. 72 item 13). Factor to be applied to catch of upper net to find the number of organisms 0.21 caught by lower net while passing through the upper stratum.

Correction Jor contamination.—The nets were lowered and raised without closing. Consequently when two nets were used, the portion of the catch of the lower net taken during its passage through the stratum fished by the upper net may be considered as a contamination. The amount of this contamination was computed from the known average concentration of mackerel eggs and larvae in the upper stratum, the known time spent by the lower net in passing through this stratum and the assumed flow through the net (the same as that registered by the current meter installed in the upper net after correction for clogging). The computations were made for each stage of egg and length of larva, and the resulting numbers subtracted from the catch of the lower net (table 14). In all instances, the corrections were substantial, and at many stations approximated the entire catch of the lower net. Important numbers usually remained after the correction at those stations where the upper net did not fish down to the thermocline and the lower net fished in the stratum above the thermocline for a time in addition to the time spent while it was being payed out and hauled back through this stratum. As might be expected from consideration of the laws of random sampling, the amounts to be subtracted were sometimes in excess of the amounts caught in the lower net. When this occurred, differences were negatively added to the catch of the upper net, these instances of over-correction offsetting other instances of under-correction, leaving the average undisturbed. Relative catch of 1 -meter and 2-meter nets. —A comparison of the catching efficiency of 1- and 2-meter nets is afforded by 19 instances during cruises VI and VII where both nets were hauled at the same station. The hauls were made, and the results were converted to the standard basis by the methods already described for both BIOLOGY OF THE ATLANTIC MACKEREL 215

nets, excepting that no current meter was employed to measure the flow of water through the 2-meter net. In lieu of this measure, the speed of towing was measured by timing the travel of the ship past a chip cast alongside. It was later found from a statistical analysis of the relation between chip speed and flow through meter nets as measured by the current meter, that tbe force of the wind modified the chip speed materially. From the relationship established, a schedide of adjustments was applied to the apparent cbip speed, to convert it to an approximation of true towing speed. This apparent flow was used instead of a current meter reading. Because of the substitution of a deduced value based in part on average performance instead of on actually measured value, the two members of individual pairs of hauls are not strictly comparable, but the average, or sum, of the 19 hauls with each type of net is not subject to this fault.

From the distribution of sizes of larvae caught by the respective nets (table 16), it is obvious that the smallest sizes of mackerel larvae were almost entirely lost through the coarse meshes of the 2- meter net; that the 6- to 9-mm. sizes were incompletely retained; and that sizes from 10 mm. upward were fully retained by the larger net. Two conclusions may be drawn from the comparison: (1) the catches of the two nets, per unit volume of water strained, are virtually identical for larvae 10 mm. and upward, and nearly so for the 7- to 9-mm. sizes, hence no material distortion can have resulted from the pooling of data from the two types of nets, according to the methods employed in this report. (2) Both types of net must have taken essentially all the larvae of sizes 10 to 22 cm. in length that chanced to be in their path, for if any larvae tended to dodge the nets they would surely have been relatively more successful in eluding the 1-meter net, and thus lowered its catch of the elusive sizes in relation to that of the 2-meter net. The closeness of the paired values for the size range specified is eloquent evidence this did not take place. It is to be regretted that no such paired hauls are available for the later cruises, when catches of still larger larvae might have indicated the upper size limit for effective catching of larvae by plankton nets.

Table 16.— Comparison of numbers of larvae caught by 1-meter nets and by 2-meter nets at identical stations of cruises VI atid VII

[Catches of both nets were converted to the basis ot straining 17.07 cubic meters per meter of depth fished]

Length of larvae (millimeters) 216 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 17.—Record of oblique havls made by 1-meter nets during cruises I to VII, inclusive, in 1932

[For explanation, see items of table 15 designated by the figures enclosed in parentheses in the column headings of this table] BIOLOGY OF THE ATLANTIC MACKEREL 217

Table 17. —Record of oblique hauls made by 1-meter nets during cruises I to VII, inclusive, in 1932- Continued 218 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 17.—Record hauls made, 1-meter nets cruises I to — of obliqve by during VII , inclusive, in 1982 Continued

Upper net Lower net

and Sta- Cruise, locality, Date Hour haul tion ' Time Flow Clog- Time Flow 1 S Clog Depth factor Depth Time ging fi r *!» , factor (4) (5) (10) (4) (5? (io7 ^? cw > (12) (7) (12) (7) (15)

CBTJIBE V

Atlantic City: L. 21440 ..do..... 885 U 21441 ..do..... 840 Ill 21442 ..do 985 IV 21443 ..do 905 Cape May: II 21447 June 4 910 Ill 21446 ..do- 690 IV.... 21445 June 3 915 V.,.., ..... 21444 ..do..... 915 Winterquarter:

I.. 21448 June 4 1, 050 II... 21449 ..do— . 890 III... 21450 ...do- 905 Chesapeake: I.. 21453 June 5 930 II 21452 ...do- 900 in 21451 June 4 910

CBtnaE vi

Martha's Vineyard: I 21468 June 8 725 II 21467 ...do— 845 III 21466 ...do- 865 Montauk: I 21464 June 7 710

II 21465 ...do— . 985

Shinnecock: II. 21463 ...do- 955 New York: II 21460 June 6 905 Ill 21461 June 7 895 IV 21462 ...do 835 Atlantic City: I...... 21459 June 6 1,075 II.... 21458 ..do— . 855 Ill 21457 ..do.... 950 Cape May: II.... 21454 June 5 905 in 21455 ...do- 855 iv 21456 June 6 855

cruise vn

Martha's Vineyard I 21490 June 19 16 10 n..._ 21491 do 20 m 21492 ..do 24 iv 21493 June 20 6 Montauk: 19 I 21489 June I 10 II 21488 ..do— 6 III.... 21487 ..do- 2 Shinnecock: I 21485 June 18 16 II- 21486 ...do 21 New York: I 21484 ...do— 7 II 21483 ..do 4 III- 21482 June 17 23 IV 21481 ...do 20 Atlantic City: I 21469 June 15 14 II 21477 June 17 6 III 21478 ...do 9 IV 21479 ...do..... 12 Cape May: III 21476 June 16 23 IV... 21475 ...do 20 V 21474 ...do 18 Winterquarter: I 21471 ...do 5 II 21472 ...do 8 BIOLOGY OF THE ATLANTIC MACKEREL 219

Table 18. —Record of oblique hauls made with a 2-meter net during cruises VIII and IX, 1932

Locality 220 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 19.—Record of mackerel eggs and larvae caught during cruises I to VII in 1932

[Numbers following the locality designation arc the serial numbers of the stations. Numbers in parenthesis are the fractions of the haul sorted for eggs and larvae. The entire haul was sorted for large larvae. The numbers given in the table are the actual counts in the sorted fractions; numbers given on the adjusted total lines are these counts converted to total catch and adjusted to represent the number per 17.07 square meter of sea surface] CRUISE I

Item BIOLOGY OF THE ATLANTIC MACKEREL 221

I to in 1932— Continued Table 19. —Record of mackerel eggs and larvae caught during cruises VII CRUISE I—Continued 222 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 19.—Record of mackerel eggs and larvae caught during cruises I to VII in 1932—Continued CRUISE II—Continued

Item BIOLOGY OF THE ATLANTIC MACKEREL 223 — Table 19. —Record of mackerel eggs and larvae caught during cruises I to VII in 1932 Continued CRUISE II—Continued 224 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 19.—Record of mackerel eggs and larvae caught during cruises I to VII in 1932— Continued CRUISE in

Item BIOLOGY OF THE ATLANTIC MACKEREL 225 — Table 19. —Record of mackerel eggs and larvae caught during cruises I to VII in 1932 Continued CRUISE III—Continued 226 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 19.—Record of mackerel eggs and larvae caught during cruises I to VII in 1932—Continued CRUISE IV

Item BIOLOGY OF THE ATLANTIC MACKEREL 227

Table 19.—Record of mackerel eggs and larvae caught during cruises I to VII in 1932—Continued CRUISE IV—Continued

Item 228 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE — Table 19.—Record of mackerel eggs and larvae caught during cruises I to VII in 193Z Continued CRUISE V BIOLOGY OF THE ATLANTIC MACKEREL 229 — Table 19. —Record of mackerel eggs and larvae caught during cruises I to VII in 1932 Continued CRUISE V— Continued 230 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 19.—Record of mackerel eggs and larvae caught during cruises I to VII in 1932— Continued CRUISE VI

Locality BIOLOGY OF THE ATLANTIC MACKEREL 231 — Table 19.—Record of mackerel eggs and larvae caught during cruises I to VII in 1982 Continued CRUISE VI—Continued 232 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 19. —Record of mackerel eggs and larvae caught during cruises I to VII in 1932— Continued CRUISE VII BIOLOGY OF THE ATLANTIC MACKEREL 233 — Table 19/—Record of mackerel eggs and larvae caught during cruises I to VII in 1983 Continued CRUISE VII—Continued 234 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE

Table 20.—Record of mackerel larvae caught on cruises VIII and IX

[Column A gives the actual count, Column B the standardized total. Sizes under 7 mm. in length have been omitted on account of their incomplete retention by the 2-meter stramin net used on this cruise] BIOLOGY OF THE ATLANTIC MACKEREL 235

Table 20.—Record of mackerel larvae caught on cruises VIII and IX— Continued CEOISE IX, JULY 16-24, 1932 236 FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE LITERATURE CITED

Bigelow, Henry Bryant. 1927. Physical oceanography of the Gulf of Maine. Bull. U. S.

Bur. Fish. vol. 40, part 2, Document No. 969, Dec. 12, 1927, pp. 511-1027, figs. 1-207. Bigelow, Henry Bryant. 1933. Studies of the waters on the continental shelf, Cape Cod to Chesapeake Bay. I. The cycle of temperature. Papers in Physical Oceanography and Meteorology, vol. 11, No. 4. Contrib. No. 34. Bigelow, Henry Bryant, and Mary Sears. 1939. Studies of the waters of the continental shelf, Cape Cod to Chesapeake Bay. III. A volumetric study of the zooplankton. Memoirs of the Museum of Comparative Zoology, vol. LIV, No. 4. Bigelow, Henry Bryant, and William W. Welsh. 1925. Fishes of the Gulf of Maine. Bull.

U. S. Bur. Fish., vol. 40, part 1, Document No. 965, pp. 1-567, figs. 1-278. Bjerkan, Paul. 1919. Results of the hydrographical observations made by Dr. Johan Hjort in the Canadian Atlantic waters during the year 1915. Dept. of the Naval Service. Canadian

Fisheries Expedition, 1914-15, pp. 349-403, 4 pis., 11 figs., Ottawa. Brice, John J. 1898. The common mackerel. In a manual of fish culture based on the methods of the United States Commission of Fish and Fisheries. Rept. U. S. Commission of Fish and Fisheries for 1897, pp. 209-213, pis. LVIII. Buchanon-Wollaston, H. J. 1935. On the component of a frequency distribution ascribable to regression. Jour, du Conseil, vol. X, No. 1, pp. 81-98. Carruthers, J. N. 1938. Fluctuations in the herrings of the East Anglian autumn fishery, the yield of the Ostend spent herring fishery, and the haddock of the North Sea in the light of relevant wind condition. Cons. Perm. Internat. p. l'explor. de la Mer, Rapp. et Proc.-Verb. CVII, 3e'me Pt. 10-15. Carruthers, J. N., and W. C. Hodgson. 1937. Similar fluctuations in the herrings of the East Anglian autumn fishery and certain physical conditions. Cons. Perm. Internat. p. l'explor. de la Mer, Rapp. et Proc.-Verb. CV, 3e'me Pt. 10-13, Copenhagen. Clark, Frances N. 1934. Maturity of the California sardine (Sardina caerulea) determined by ova diameter measurements. Calif. Div. Fish and Game. Fish Bull. 42; 49 pp., 19 figs. Dannevig, Alf. 1919. Canadian fish-eggs and larvae. Canadian Fisheries Expedition, 1914-15, pp. 1-74, 3 pis., 26 figs., Dept. of Naval Service, Ottawa. Ehrenbaum, E. 1921. Ueber die Makrele. Wissensch. Meeresuntersuchungen. Abt. Helgo- land. Bd. XV. No. 2, 1921. Embody, George Charles. 1910. The ecology, habits and growth of the pike, Esox lucius. Unpublished thesis deposited in the library of Cornell University. Ezekiel, Mordecai. 1930. Methods of correlation analysis. John Wiley & Sons, Inc., New York, 1930. Fiedler, Reginald Hobson. 1932. Fishery industries of the United States, 1931. Report of U. S. Commissioner of Fisheries for the fiscal year 1932, App. II. Oct. 12, 1932, pp. 97-440. Fiedler, Reginald Hobson, John Ruel Manning, and Fred Francis Johnson. 1934. Fishery industries of the United States, 1933. Report of U. S. Commissioner of Fisheries for the fiscal year 1934, App. I. Aug. 17, 1934, pp. 1-237. Fish, Charles John. 1928. Production and distribution of cod eggs in Massachusetts Bay in 1924 and 1925. Bull. U. S. Bur. Fish., vol. 43, pt. 2. Document No. 1032, Feb. 25, 1928, pp. 253-296, figs. 1-16. Fisher, R. A. 1932. Statistical methods for research workers. Oliver and Boyd, Edinburgh. Goode, George Brown. 1884. The mackerel (Scomber scombrus) in Fishery Industries of the

United States. Natural Hist, of Useful Aquatic Animals. Sect. 1, pp. 281-303, U. S. Govt. Printing Office. Garstang, Walter. 1898. On the variation, races, and migrations of the mackerel, Scomber scombrus. Jour. Marine Biol. Assoc, vol. V, No. 3, pp. 235-295. Hjort, Johan. 1926. Fluctuations in the year class of important food fishes. Jour, du Conseil,

Feb. 1926, vol. I, No. 1, pp. 5-39. Huntsman, A. G. 1919. Some quantitative and qualitative plankton studies of the Eastern Canadian plankton. Canadian Fisheries Expedition, 1914-15, pp. 405^85, 12 figs., Dept. of Naval Service, Ottawa. Iselin, Columbus O'Donnell. 1936. A study of the circulation of the western N. Atlantic. Papers in Physical Oceanography and Meteorology. Vol. IV, No. 4, August 1936. BIOLOGY OF THE ATLANTIC MACKEREL 237

Marshall, S. M., A. G. Nicholls, and A. P. Orr. 1937. On the growth and feeding of the larval and post-larval stages of the Clyde herring. Jour. Marine Biol. Assoc, vol. XXI, No. 1, Nov. 1937, pp. 245-267. Moore, J. Percy. 1899. Report on mackerel investigations in 1897. Report Commissioner, U.

S. Comm. Fish and Fisheries for 1898, [App. I.,] pp. 1-22. Russell, F. S. 1925. Depth recording with plankton-nets. Nature, vol. CXV, 1925, pp. 603- 604. London. Sette, Oscar Elton. 1931. Outlook for the mackerel fishery in 1931. Fishery Circular No. 4, Bur. of Fish., U. S. Dept. Commerce, Aug. 1931, pp. 1-20. Sette, Oscar Elton. 1932. Outlook for the mackerel fishery in 1932. Fishery Circular No. 10, Bur. of Fish., U. S. Dept. Commerce, June 1932, pp. 1-25. Sette, Oscar Elton. 1933. Outlook for the mackerel fishery in 1933. Fishery Circular No. 14, Bur. of Fish., U. S. Dept. Commerce, May 1933, pp. 1-23. Sette, Oscar Elton. 1934. Outlook for the mackerel fishery, 1934. Fishery Circular No. 17, Bur. of Fish., U. S. Dept. Commerce, April 1934, pp. 1-6. Sette, Oscar Elton. 1938. Relative strength of year classes of certain fishes of the Western Atlantic. E. Mackerel. Cons. Perm. Internat. p. l'explor. de la mer, Rapp. et Proc.-Verb.

CI, 4e'me, Pt. 1, Copenhagen. Sette, Oscar Elton, and A. W. H. Needler. 1934. Statistics of the mackerel fishery off the east coast of North America, 1804 to 1930. Inv. Rept. No. 19, Bur. of Fish., U. S. Dept. of Commerce, 1934, pp. 1-48. Sparks, M. Irving. 1929. The spawning and development of mackerel on the outer coasl of Nova Scotia. Contrib. to Canad. Biol, and Fisheries, N. S., vol. IV, No. 28, pp. 443-452. Walford, Lionel Albert. 1938. Effect of currents on distribution and survival of the eggs and larvae of the haddock (Melanogrammus aeglefinus) on Georges Bank. Bull. U. S. Bur. Fish., vol. 49, Bull. No. 29, pp. 1-73, figs. 1-50. Wilson, Henry V. P. 1891. The embryology of the sea bass (Serranus atrarius). Bull. U. S. Fish. Comm., vol. 9, Document No. 160, July 7, 1891, pp. 209-277, pi. LXXXVIII-CVII, text-figs. 1-12. Winsor, Charles P., and Lionel A. Walford. 1936. Sampling variations in the use of plankton nets. Jour, du Conseil, vol. XI, No. 2, pp. 190-204. Worley, Leonard G. 1933. Development of the egg of the mackerel at different constant temperatures. Jour. Gen. Physiology, vol. 16, No. 5, pp. 841-857. o